• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Flowrate...

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
BillA -

Two questions...

1) What's a MCW 462U?
2) Where do I send the Glen Livet?


I don't have any fancy gear so my setup is going to be plain jane. I intend to use tap water to cool the rig. I'll measure the water temp but I'm reasonably confident that tap water temperature won't vary during the run. I'll use another thermometer to measure the outlet water temperature. The weakest part of all this will be I'll be using a mobo sensor for my cpu temp. I don't have an xp with an onboard diode. I figure that though the cpu temp measurement will be ideosyncratic with respect to my mobo, what I'm looking for is to see where the temp curve flattens out so any error introduced by the mobo sensor will be constant across water flows. Also, the mobo temp curve should track with the outlet temperature which I will measure directly.

I'll use my siphon rig to vary the water flow. More flow will be achieved by raising the source bucket. I'll maintain a constant head of pressure by ensuring the inlet bucket is overflowing and the outlet stays in the same spot while the test is underway.

So I think by using tap water I'll be simulating a perfect radiator that always dumps every btu coming its way. By measuring the outlet water temp, I'll have an accurate measure of how much heat the water block shed and by monitoring the mobo temp, I'll be confident that the cpu isn't on its way to perdition. By using a simple siphon to control water flow, I can vary the gpm without having to buy a new water pump that's capable of delivering Niagra Falls.

Comments?
 
gone_fishin said:
Other things in the system could limit the waterflow like the block itself. Which block are you going to use?
<Robert De Niro voice>You talkin' to me?</Robert De Niro voice> I'll be using a Gemini. You're right it may serve as a throttle but we'll see what happens. Hopefully, it'll let 2 gpm through.
 
Re: Re: Re: Re: Re: Re: Re: Answer for Neo86

parapapa? said:
Friction is going to change flow rate, yes I know this, but very, very minimal. Not even enough to really argue about. "The ONLY point I was trying to make to Neo was that the decrease in volume of the water increased his temperatures, not the shortening of hoses (that is MY OPINION).
GreenmanWD-40
What you're failing to grasp is BillA hostility was engendered because you presented your opinion as fact. There have been a lot of posts to this thread that are nonsense and unfortunately, your post was one of them.

For example, take the tail of your first sentence "very,very minimal." You and I have different thresholds for minimal. The difference I measured was 32 seconds for 32" vs 44 seconds for 100" of tubing. The shorter tube beat the longer tube by 27%. Is a rig going to have 100" of tubing. Yep, I'm building one right now.

You're also going to get nailed because you're saying that temperature is a function of a system's water volume. You've been told more than once now that that is wrong. In the situation that you were responding to, the lost volume due to shorter tubing isn't going to make any difference in the rig's equilibrium temperature. It'll affect the time it takes to reache equilibrium but it won't affect the final steady-state temp.

As long as you spout off nonsensical opinions, you're going to get nailed. [/B][/QUOTE]



NOW,


We are talking about 6 to 16 inches of tubing not 32 or 100. I continue to say, very minimal. I'll tell you what, I'll do some controlled experiments and post the data (when I have more free time). I'll take back what I said about very minimal if it exceeds 10 or 15%.

LET ME SAY AGAIN, MY ORIGINAL OPINION WAS THAT NEO'S TEMPERATURE GAIN (VERY FIRST THREAD) WAS "MAINLY" FROM THE COOLING CAPACITY OF THE WATER MISSING AND NOT THE FLOWRATE.

GreenmanWD-40
 
Re: Re: Re: Re: Re: Re: Re: Re: Answer for Neo86

GreenmanWD-40 said:

GreenmanWD-40
What you're failing to grasp is BillA hostility was engendered because you presented your opinion as fact. There have been a lot of posts to this thread that are nonsense and unfortunately, your post was one of them.

For example, take the tail of your first sentence "very,very minimal." You and I have different thresholds for minimal. The difference I measured was 32 seconds for 32" vs 44 seconds for 100" of tubing. The shorter tube beat the longer tube by 27%. Is a rig going to have 100" of tubing. Yep, I'm building one right now.

You're also going to get nailed because you're saying that temperature is a function of a system's water volume. You've been told more than once now that that is wrong. In the situation that you were responding to, the lost volume due to shorter tubing isn't going to make any difference in the rig's equilibrium temperature. It'll affect the time it takes to reache equilibrium but it won't affect the final steady-state temp.

As long as you spout off nonsensical opinions, you're going to get nailed.



NOW,


We are talking about 6 to 16 inches for most systems of tubing not 32 or 100. I continue to say, very minimal. I'll tell you what, I'll do some controlled experiments and post the data (when I have more free time). I'll take back what I said about very minimal if it exceeds 10 or 15%.

LET ME SAY AGAIN, MY ORIGINAL OPINION WAS THAT NEO'S TEMPERATURE GAIN (VERY FIRST THREAD) WAS "MAINLY" FROM THE COOLING CAPACITY OF THE WATER MISSING AND NOT THE FLOWRATE.

GreenmanWD-40 [/B][/QUOTE]
 
BillA said:
here is a more intelligible plot

462U%2070W.jpg


again, at 70W with a heat die

but do appreciate that these curves are generated using a lab chiller to maintain the coolant inlet temp at 25.00°C, +/-0.01°

with the typical WCing system the coolant temp will rise with the flow rate as the radiator will not be able to do as 'good' a job

be more than willing to offer suggestions on your proposed test setup
the more people generating real numbers, the less hocus-pocus there will be
such measurements become a challenge to achieve the needed accuracy and resolution

the above curve is 'smooth' because the temps were recorded in hundredths, rounded in tenths its nowhere near so nice

be cool

What's the value at 4 gpm?

GreenmanWD-40
 
This post deleted by its "derogatory and insulting" author so that it will not be (so) necessary
"to defend the little guys against the tyranny of the know-it-alls."
 
Last edited:
This post deleted by its "derogatory and insulting" author so that it will not be (so) necessary
"to defend the little guys against the tyranny of the know-it-alls."
 
Last edited:
BillA said:
HEY parapapa?

I almost forgot; and the Krispy Cremes ??
(its ok, they don't ship well)

be cool
Boy! It's a good thing you didn't lose the bet! You were supposed to send me a Krispy Kreme if I won. I was giving you odds.

Nonetheless, send me your address and a bottle of 12 year old will be on its way.
 
This post deleted by its "derogatory and insulting" author so that it will not be (so) necessary
"to defend the little guys against the tyranny of the know-it-alls."
 
Last edited:
I'm still serious also. I have looked at the plans from mldhab for building the simultated cpu heat load and the guy doing all my machine work says he can make it for me.

Still a few other things to get out of the way but getting there.

Also I understand the need to completely isolate a specific component and test a single factor for that component only but is this level of controle really needed if the desired result is to test a complete setup as a whole?

What I'm getting at is that if I want to test complete systems to determine the actual cooling efficiency of each system then the need is not to find the weakest link on a system or to determine why it works like it does but to show that setup a works beter than setup b. The only reason to goto that level would be to test my own equipment to find the weakest link.

Thanks BillA.
 
BillA said:
vandersl

re the cooling actually achieved by a rad:
several different measures can be used to gauge a radiator's 'performance'
- one is the heat dissipation calculated in the normal way
- another is the coolant temperature reduction across the rad

these are in contradiction
maximum dissipation is (normally) at maximum flow
maximum temperature drop is at a (somewhat undefined) 'minimum' flow

in the radiator article the rads were tested, and rated, as components

in a WCing system however they work interactively with the pump and waterblock to induce an effect in another component; lower temperatures in the CPU via the TIM joint

so,
as the flow rate increases the heat convection rate in the wb increases (which 'drives' the CPU temp down)
and,
the temp rise across the wb decreases
thus
the rad is then receiving coolant at a lower temp (as compared to the slower flow rate) which reduces its efficiency
while at the same time the temp drop across the rad is decreased due to the higher flow rate

the net effect seems to be that if there is substantial overcapacity in the rad, then the system can benefit from higher flow rates

there are some important caveats:
- the system must be designed for minimal flow restriction everywhere except the wb
- a pump must be used that has a higher head capability AND that does not input large amounts of heat into the coolant
(very few pumps are of this type)
- the wb must be able to translate higher flow into increased cooling, which is to say that the incurred head loss in the wb must result in a higher heat convection rate
(an extreme bad example would be a 2 in. dia. short pipe section with a flat on its side to be pressed against the CPU; low flow restriction and poor convection)

this is not how most WCing systems are 'designed' to operate, for good reason;
such systems are not efficient in terms of cost, noise, space, or just about any other measure
the 'law of diminishing returns' will not be denied

garasaki, thanks

parapapa?

the MCW462-U is Swiftech's 'big block', though not the big barb version
serious ? my address is on the site hosting the images

- tap water will work, but have NO aluminum in the system, it will start 'blooming' within an hour
- were I you, I would defer the project until I had a mobo and CPU to utilize a Maxim reader
(see the article and threads by pHaestus re all this - I do not use a CPU as a heat source and am HIGHLY critical of the 'smoke' Watts so generated, and the always unaccounted secondary path losses)
"More flow will be achieved by raising the source bucket." - gonna be a real TALL ladder to get the changes you're looking for, do the calcs
(buy a Little Giant pump, $30 on eBay, and use a dump valve - do not throttle)

your problem (as if there were no others, < g >) is with the coolant temperature measurements
can you make accurate measurements to at least 0.1°C ?
if you cannot, give it up
(buy a Fluke 2190A and Y2001 off eBay, often $50 !!)

GreenmanWD-40

you are a truly hopeless case, got your balls up in front of your eyes and can't see past 'em
Darcy-Weisbach formula, Fanning formula,
First Law of Thermodynamics for Pipe Flow; Head Loss

I have no interest in your ability to replicate testing done in student hydraulics labs the world over
can you not read ?
can you not integrate the information you have (presumably) read ?
you are flogging a horse having rigor mortis

be cool

BillA

Apparently you are a complete moron in addition of being illiterate. The reason for this is because I only was giving my opinion to Neo (you know the first thread), about the reason for the temperature gain. Then you spoke up about your “fluid friction factor”, WHICH IS COMPLETE BS AND WOULDN'T APPLY THERE. Your right in most of what you say, but apparently you cannot read. For someone who is probably 40 years of age and plus (assuming your not a liar), you act 13, maybe 15 in the maturity level.

Maybe you will agree to disagree? Since it's my opinion.

Maybe the best reply to “god” :rolleyes: is no reply at all?
 
This has been a good discussion minus the heat:argue:

BillA, to quote you and ask after
Quote
"there are some important caveats:
- the system must be designed for minimal flow restriction everywhere except the wb"
end quote

This is the exact opposite of my working example in my system, as it pertains to the waterblock, yet my system performs up to standards. How do you account for this?
 
GreenmanWD-40 said:

Maybe you will agree to disagree? Since it's my opinion.


To be diplomatic with this everyone has an opinion. The only thing that matters is if you can back it up with hard data. What bill and others have been trying to tell you is more or less put your money were your mouth is.

Prove it.

Me personnaly I've seen so much garbage floating around here and other forums that I simply started testing everything for myself. I'm not an engineer of any kind. I actually specilized in system admin. My background is more in programing but at least I have enough math to follow part of whats being said. To tell the truth I don't even care about what can be proved by math. It's good and all but unless ther are numbers to back it up it's nothing but somebodies mathimatical fantasy.

So I say to you enough of opinions and prove it. If I can with my limited knowledge of thermodynamics (it's been so many years since my math courses I was good at it but don't use most of it anymore) take some of this and be able to aply it and even test for it, then you with the background should be able to outproduce me any day of the week. Take at least this one chance think back to the days when you used to be in class and do it right from begining to finish. Don't accept anything from yourself that your prof wouldn't. Now take this opertunity and learn from it. It matters not who is right just do the experiments and prove it.


You may be right and if so will be able to prove it. You may be wrong and if so you will learn from it.
 
Last edited:
gone_fishin said:
BillA, to quote you and ask after
Quote
"there are some important caveats:
- the system must be designed for minimal flow restriction everywhere except the wb"
end quote

This is the exact opposite of my working example in my system, as it pertains to the waterblock, yet my system performs up to standards. How do you account for this?

exact opposite ? not sure I follow
your overall system seems to have low flow resistance
your pump is quite suitable
your wb seems to cool well (how well relatively we cannot determine from what is known)

note the Swiftech 462
there are many ways to skin a cat

GreenmanWD-40

how about a 'deal' ?
I'll cease this ugly pounding on you;
and you read - with an open mind - some of the references I've given
???

you have every right to an opinion,
but you must also accept that if such can be PROVEN wrong, then it will be called such
and if you are then unable (or unwilling) to understand that your opinion has been shown to be wrong, in this specific case by both theory and experiment, then those appellations applicable to such small close-minded people will be applied to you in a very personal manner
- for as long as you choose to assert the incorrect opinion
(and with me this assessment/appellation will transpire in a single pair of transactions)

is it really so hard to say "hey, I was wrong" ?
I, for one, do make errors all the f***ing time, humble pie is a staple on my plate
and so what ? this acceptance enables me to learn form my errors
every day is a new lesson; the trick is if we are able to recognize, and accept, the lesson
do learn to accept the lesson at once, else it WILL be back tomorrow, and the next day, . . .

be cool
 
BillA,

The following is not an argument to your post, per se, but a follow-up that may be of help to anyone reading and actually caring about the theory behind such things.

re the cooling actually achieved by a rad:
several different measures can be used to gauge a radiator's 'performance'
- one is the heat dissipation calculated in the normal way
- another is the coolant temperature reduction across the rad

These are actually the same - the heat dissipated by the rad will equal the heat lost by the coolant.

these are in contradiction
maximum dissipation is (normally) at maximum flow
maximum temperature drop is at a (somewhat undefined) 'minimum' flow

Exactly - while the two measures are the same, optimizing for either is not. In particular, optimizing for maximum coolant temperature drop will not result in maximum heat dissipation.

in the radiator article the rads were tested, and rated, as components

I have some questions on your radiator testing article (not for here) - I still maintain that the idea that heat dissipation can go down with increasing flow rates is counter to logic (though the idea that heat dissipation will increase significantly with higher flow rates is also somewhat counter to logic). In fact, not one of the performance graphs for the rads on the Lytron site you reference in the article show this phenomenon.

in a WCing system however they work interactively with the pump and waterblock to induce an effect in another component; lower temperatures in the CPU via the TIM joint

This is true - awareness of the system as a whole is important, if only to be able to pinpoint the area of the system which would most benefit from optimization.

so,
as the flow rate increases the heat convection rate in the wb increases (which 'drives' the CPU temp down)
and,
the temp rise across the wb decreases
thus
the rad is then receiving coolant at a lower temp (as compared to the slower flow rate) which reduces its efficiency
while at the same time the temp drop across the rad is decreased due to the higher flow rate

This is where the math throws a kink in, in terms of magnitude. With a 100W CPU load, here are the temperature increases in WB/drops in rad of water for various flow rates:
0.5gpm 0.76C
1.0gpm 0.38C
2.0gpm 0.19C
3.0gpm 0.14C

Considering that most WC setups have coolant temperatures several degrees above ambient, changes in coolant temperature in the rad for various flow rates should change the efficiency very little (for this load, anyway). If the coolant were on average 3C above ambient, then the change from 1.0gpm to 2.0gpm will reduce the average radiator coolant temp from 3.19C to 3.08C, or 0.09C. Whether or not this outweighs the system benefits from the increased flow rate is another story.

the net effect seems to be that if there is substantial overcapacity in the rad, then the system can benefit from higher flow rates

Hmmm - haven't put much thought into this, but I think I could argue the opposite. If the internal surface area of the rad is insufficient for the heat load, then there will be a larger delta-T between the coolant and the rad walls. In this case, increasing flow rates would reduce the required delta-T (by increasing the heat transfer coefficient). Note that this is done without changing the net heat load on the radiator itself - 100W is still 100W. In the end, if the radiator fins need to be at temperature X to dissipate the heat, then increasing the flow rate will decrease the temperature of the fluid required to keep the walls/fins at temp X.

there are some important caveats:
- the system must be designed for minimal flow restriction everywhere except the wb
- a pump must be used that has a higher head capability AND that does not input large amounts of heat into the coolant
(very few pumps are of this type)
- the wb must be able to translate higher flow into increased cooling, which is to say that the incurred head loss in the wb must result in a higher heat convection rate
(an extreme bad example would be a 2 in. dia. short pipe section with a flat on its side to be pressed against the CPU; low flow restriction and poor convection)

These are all important caveats. Increasing pump power is not the way to increase flow beneficially. With a 100W CPU, going from a 10W pump to a 45W pump is a substantial increase in power (not sure how much goes into the water - depends). But increasing flow by shortening hoses, increasing connectors sizes, and reducing the use of elbows is essentially 'free'.

this is not how most WCing systems are 'designed' to operate, for good reason;
such systems are not efficient in terms of cost, noise, space, or just about any other measure
the 'law of diminishing returns' will not be denied

Exactly - seems with most systems, the biggest benefit to be gained is by increasing air flow through the rad - but that comes at the expense of noise. C'est la vie.

Just for info - my educational/professional background is in EE. I do not claim to be a fluidic/thermodynamic expert. So feel free to criticize anything I have wrong. I have no personal stake in this - just trying to put perspective on an issue central to WC.

Be cooler :)
 
BillA said:


exact opposite ? not sure I follow
your overall system seems to have low flow resistance
your pump is quite suitable
your wb seems to cool well (how well relatively we cannot determine from what is known)

note the Swiftech 462
there are many ways to skin a cat

be cool

Perhaps I just worded the question wrong, I only meant the waterblock, as my method has been with minimal resistance there. I'll have to look up the swiftech to see what you mean.
Thanks for the input, give em hell:beer:
 
vandersl

thanks for the initial astute question,
and the even better insightful and explanatory commentary

an observation about myself:
I am a test results oriented tinkerer, not a theoretician by any means
but valid test results MUST match the theory, or the reasons for the deviation understood
over the last 2 years I've been taking daily lessons in thermodynamic testing, and I am finding it very difficult to do well
in particular low amounts of heat at low flow rates are quite prone to error - as might be expected

my generalizations about the import of the test results are most often a synthesis of other's views garnered from as wide a range of sources as I can find, and all such kind of need to agree

wrt the Lytron product data:
note that they describe a 'system', rather than a component in isolation; and that their flow rates are considerably higher along with the delta Ts
this is how industrial applications 'work', only WCers try to cool back to ambient

for sure heat in MUST equal the heat out
and this applies to the wb and rad individually, and to the system as a whole
but the thermal balance calculations necessitate an EXTREME degree of accuracy and precision for the coolant temps in the heat ranges we are considering

a WCing system will always balance itself, and this can be viewed in terms of the coolant temp - or the CPU temp;
both under steady-state conditions of course

I have not yet encountered any condition of higher coolant temps that result in lower CPU temps
so I tend to be more focused on rads than wbs (and indeed most 'current' wb seem very similar, with several exceptions perhaps a couple of degrees lower)

but I have a slew of high-pressure pumps awaiting a suitable wb

be cool
 
BillA said:
vandersl

thanks for the initial astute question,
and the even better insightful and explanatory commentary

an observation about myself:
I am a test results oriented tinkerer, not a theoretician by any means
but valid test results MUST match the theory, or the reasons for the deviation understood
over the last 2 years I've been taking daily lessons in thermodynamic testing, and I am finding it very difficult to do well
in particular low amounts of heat at low flow rates are quite prone to error - as might be expected

my generalizations about the import of the test results are most often a synthesis of other's views garnered from as wide a range of sources as I can find, and all such kind of need to agree

wrt the Lytron product data:
note that they describe a 'system', rather than a component in isolation; and that their flow rates are considerably higher along with the delta Ts
this is how industrial applications 'work', only WCers try to cool back to ambient

for sure heat in MUST equal the heat out
and this applies to the wb and rad individually, and to the system as a whole
but the thermal balance calculations necessitate an EXTREME degree of accuracy and precision for the coolant temps in the heat ranges we are considering

a WCing system will always balance itself, and this can be viewed in terms of the coolant temp - or the CPU temp;
both under steady-state conditions of course

I have not yet encountered any condition of higher coolant temps that result in lower CPU temps
so I tend to be more focused on rads than wbs (and indeed most 'current' wb seem very similar, with several exceptions perhaps a couple of degrees lower)

but I have a slew of high-pressure pumps awaiting a suitable wb

be cool

Now,

How about this.

From the first thread, you think, or know, the main cause of the temperature increase is just the friction (which will decrease the flowrate)?

Here's one off the kind of off the subject.

90 degree bends need to be place before and after the heat xchanger and the waterblock to enhance trubulent flow, would you agree with that? Also turbulent flow will enhance heat transfer?
 
I'll answer these out of order.

Also turbulent flow will enhance heat transfer?

Absolutely agree. Turbulent flow enhances heat transfer. Well documented, fits the theory. Got that one right.

90 degree bends need to be place before and after the heat xchanger and the waterblock to enhance trubulent flow, would you agree with that?

Absolutely disagree. Especially with the 'after' the exchanger and waterblock, where introducing turbulence would have no real benefit.

You can get turbulent flow in a straight tube with smooth walls if the water flows fast enough (usually within the range of fluid velocities used in WC). This is the result of fluid viscosity and velocity differential between the fluid at the tube walls and the center of the channel. I hesitate to bring up 'fluid friction' again in this discussion, but there it is :)

Spikes, swirls, 'tubulators' and other flow channel designs to induce turbulence may not be needed in many cases (but they look cool, and sell more blocks).

I think the reason most blocks use inlets at 90 degrees to the CPU surface is that its hard to get clearance around the socket for the barbs if they are parallel to the CPU surface.

Injecting the coolant from above the block surface is a reasonable way to induce turbulence as well (a la the swiftech block), without restricting flow needlessly.
 
Back