Pump->CPU or Pump->HC->CPU ?(other aspect, NOT heat)

OTMOPO3OK · 03-20-2003, 04:58 PM

I know this question has been asked before but all the discussions revolve about pump adding heat to the system and that it's not that much(0.5C) at the speed the water is flowing..
My question is if i run Pump->HC->CPU wouldn't HC defeat the pressure the pump puts out...
if the water would be going straing to CPU at maximum velocity...i think if it goes thru HC it already has some restriction till it gets to the CPU...
but then again the speed in a closed loop should be the same at most parts...i'm confused..
i mean somewhere in the back of my mind i think that running Pump->CPU is better....since it's less restriction...
Any comments on that...

airspirit · 03-20-2003, 05:14 PM

Screw your pump adding heat:

rad->res->pump->block

This allows a fresh and unrestricted water supply to hit the pump, and protects against hose collapse. Plus, the cooler the water to your pump is, the longer the pump will last.

I've lost a CPU/Mobo to hose collapse, so I know.

jtroutma · 03-20-2003, 06:15 PM

I agree with the arrangement: Pump->Block->Rad->Res.

The highest velocity of water will be comming straight off the pump. Also having the Res. before the pump lowers the intake resistance to the pump. Lastly, having the radiator between the radiator and pump will not help all that much if at all. (cooling the water just after it was heated up makes more sence to me)

This is how I have my system setup.

Mark Larson · 03-20-2003, 06:18 PM

Hose collapse? Try tubing that's twice as thick as normal tubing - 0.25" instead of 0.125". (You'll want ¾" OD and ½" ID)

BTW, good thread. I've been wondering about this myself, especially wrt pressure.

Alchemy · 03-20-2003, 08:19 PM

OTMOPO3OK is mostly right. So long as your tubing is the same diameter, the velocity of the water is going to be the same at any point in the tubing. Likewise, the pressure drop across the HC and across the WB will be the same regardless of what order they are in. So as far as heat transfer properties go, they will be the same no matter what order they are in.

More important factors are how much flow restriction is caused by bends in the tubing. Avoid those.

There's heat input due to the pump is often overrated, I think, except in cases where people are using huge pumps near the heat output of the processors they're trying to cool. Put the pump before the rad if you know it won't cause more problems with your piping system. Otherwise, don't worry about it.

Alchemy

MadDogMe · 03-21-2003, 03:43 AM

I think the argument was that it's more 'turbulant' straight from the pump. I'd use the shortest tube lenght possible between pump and block, connect it straight on if possible...

I'd of thought the flow would be equal everywhere but pressure would be lost after/due~to each restriction on route?, thereby the pressure would be greater at the block with a pump>CPU arrangment. water pressure helps cooling no?, more impingement?...

chesspatzer · 03-21-2003, 07:18 AM

A Simplified Generalization:

If I'm applying the flow formula correctly: m=pAV, where m is flow rate, p is 'rho' which is the density of water (ignore in this case), A is cross sectional area of the tube, and V is velocity.

Flow is constant throughout the system, that is, you don't have a greater amount of water flowing out of the pump than you have flowing in to it and the same for any component. Since the flow is constant and the density is (basically) constant, the above formula can be reduced to AV=c so that the Area times the Velocity equals some constant. Therefore, as the area increases, the velocity decreases and vice versa.

The point is this, the arrangement of components, Pump->HC->CPU or Pump->CPU->HC, is irrelevant on velocity. It is the cross sectional area of the Block the governs velocity.

It's been a little while since I've used my engineering background but here's what I think really matters. Since you want to draw heat off of the CPU, a "convective heat transfer process", the basic formula is: Q=hA(dT), where Q is the rate of heat transfer (bigger is better), h and A are dependent on the block characteristics, and (dT) is delta T or temperature difference. The only thing that you can control by your setup is the delta T. The bigger the delta T, the better. Now, what is delta T? It is the temperature difference between the surface temperature of the CPU block and the water flowing across it. Specifically it is t(s) minus t(w) where t(s) is the temperature of the surface and t(w) is the temperature of the water. To make this number bigger, our goal, we can make the CPU hotter, not what we want, or make the water cooler!

Therefore, in my old, rusty, dusty, mind - Pump->HC->CPU would be preferred because it puts the lowest possible water temperature into contact with the CPU.

winewood · 03-21-2003, 07:31 AM

Kudos for keeping it simple. Thank you. I believe you are one of the few that don't explode my head with characters and equations that arent confusing. (baby steps)
I believe from the discussions I read before, the thing that goes down in a loop from pump to end unit is pressure. So.. flow rate stays the same, but the pressure drops as it flows away from the source. Keeping the pump -> block the psi will be highest, at this point, and that can help with the transfer of heat. The radiator is one of the larger contributors to psi drop, therefore keeping it after the block contributes to a larger psi at block. The before mentioned configuration sounds logical to me, and I use it myself.
Did I totally botch that understanding? Corrections welcome.

OTMOPO3OK · 03-21-2003, 10:03 AM

thx for replies guys...
chesspatzer, you too for explanation
but i have to agree with winewood i still feel that putting CPU after pump gives it more turbulance ( i use nozzle)...
i wish i could test both but i have very limited supply of tubing i got from my Chemistry lab professor and don't want to waste it trying out things...maybe i'll get some cheap silicone and try that...

Cova · 03-21-2003, 10:18 AM

I'm pretty sure I read on these forums somewhere by one of the other people who can never have enough math in a post (I believe it was BillA) a different forumula for calculating the heat transfer from the block to the coolant, and that pressure was involved in it. It's not as simple as a convective heat transfer - it would be convection getting the heat in the coolant that actually touches (at the molecular level) the block to the rest of the coolant flowing through, but it is conduction getting the heat from the edge-molecules of the copper into whatever coolant molecules come into contact with it.

Either way, I don't think it makes a measureable difference (just like I don't think pump-heat is a measurable enough difference to argue about where it should go), and recommend that the pump is placed either for aesthetic purposes in the case of big windows, or wherever simplifies tube-routing the most for cooling purposes. As for pump-created turbulence, I don't think that will make a measurable difference either unless you pretty much just put the pump-output right on/in the block (as a few here have tried).

OTMOPO3OK · 03-21-2003, 11:10 AM

Quote:

Originally posted by Cova
put the pump-output right on/in the block

Holly cow... that's crazy...that's whole lots of trouble mouting the pump right on block...i think a little tubing between them won't change anything...this is crazy...

Alchemy · 03-30-2003, 06:25 PM

Quote:

Originally posted by chesspatzer
It's been a little while since I've used my engineering background but here's what I think really matters. Since you want to draw heat off of the CPU, a "convective heat transfer process", the basic formula is: Q=hA(dT), where Q is the rate of heat transfer (bigger is better), h and A are dependent on the block characteristics, and (dT) is delta T or temperature difference. The only thing that you can control by your setup is the delta T. The bigger the delta T, the better. Now, what is delta T? It is the temperature difference between the surface temperature of the CPU block and the water flowing across it. Specifically it is t(s) minus t(w) where t(s) is the temperature of the surface and t(w) is the temperature of the water. To make this number bigger, our goal, we can make the CPU hotter, not what we want, or make the water cooler!

Not quite. Q=hA(dT). Heat output (Q) is constant - there's nothing you can do to the WC system to make the CPU put out more heat. hA is a function of the block, the water in it, and how they interact. You can improve hA by increasing flow rate and improving the WB design.

Thus dT is constant. Applying colder water will not change dT. It will decrease the CPU temperature to keep dT constant.

Same result - colder water temp is good.

Alchemy

Skulemate · 03-30-2003, 07:26 PM

Quote:

Originally posted by chesspatzer
If I'm applying the flow formula correctly: m=pAV, where m is flow rate, p is 'rho' which is the density of water (ignore in this case), A is cross sectional area of the tube, and V is velocity.

You can write the continuity equation without density as well... Q=AV. This will give you flow in terms of L^3/s instead of M/s.

03-20-2003, 04:58 PM	#1
OTMOPO3OK Cooling Neophyte Join Date: May 2002 Location: Brooklyn, NY Posts: 45	Pump->CPU or Pump->HC->CPU ?(other aspect, NOT heat) I know this question has been asked before but all the discussions revolve about pump adding heat to the system and that it's not that much(0.5C) at the speed the water is flowing.. My question is if i run Pump->HC->CPU wouldn't HC defeat the pressure the pump puts out... if the water would be going straing to CPU at maximum velocity...i think if it goes thru HC it already has some restriction till it gets to the CPU... but then again the speed in a closed loop should be the same at most parts...i'm confused.. i mean somewhere in the back of my mind i think that running Pump->CPU is better....since it's less restriction... Any comments on that... __________________ Keyboard not found. Press F1 to continue...

03-20-2003, 05:14 PM	#2
airspirit Been /.'d... have you? Join Date: Jul 2002 Location: Moscow, ID Posts: 1,986	Screw your pump adding heat: rad->res->pump->block This allows a fresh and unrestricted water supply to hit the pump, and protects against hose collapse. Plus, the cooler the water to your pump is, the longer the pump will last. I've lost a CPU/Mobo to hose collapse, so I know. __________________ #!/bin/sh {who;} {last;} {pause;} {grep;} {touch;} {unzip;} mount /dev/girl -t {wet;} {fsck;} {fsck;} {fsck;} {fsck;} echo yes yes yes {yes;} umount {/dev/girl;zip;} rm -rf {wet.spot;} {sleep;} finger: permission denied

03-20-2003, 06:15 PM	#3
jtroutma Cooling Savant Join Date: Aug 2001 Location: SLO, CA Posts: 837	I agree with the arrangement: Pump->Block->Rad->Res. The highest velocity of water will be comming straight off the pump. Also having the Res. before the pump lowers the intake resistance to the pump. Lastly, having the radiator between the radiator and pump will not help all that much if at all. (cooling the water just after it was heated up makes more sence to me) This is how I have my system setup. __________________ Athlon64 X2 4200+ @ 2.5Ghz (250FSB x 10) OCZ VX 1GB 4000 @ 250FSB (6-2-2-2 timmings) DFI LANParty nForce4 Ultra-D SCSI Raid 5 x (3) Cheetah 15K HDDs LSI Express 500 (128MB cache) OCZ PowerStream 520W PSU ATI X850XT PE (Stock) DTEK WhiteWater + DTEK Custom Radiator Eheim 1250

03-20-2003, 06:18 PM	#4
Mark Larson Cooling Neophyte Join Date: Feb 2003 Location: MN Posts: 67	Hose collapse? Try tubing that's twice as thick as normal tubing - 0.25" instead of 0.125". (You'll want ¾" OD and ½" ID) BTW, good thread. I've been wondering about this myself, especially wrt pressure. __________________ AMD Fanboy.

03-21-2003, 03:43 AM	#6
MadDogMe Thermophile Join Date: Aug 2002 Location: Just shut up ;) ... Posts: 1,068	I think the argument was that it's more 'turbulant' straight from the pump. I'd use the shortest tube lenght possible between pump and block, connect it straight on if possible... I'd of thought the flow would be equal everywhere but pressure would be lost after/due~to each restriction on route?, thereby the pressure would be greater at the block with a pump>CPU arrangment. water pressure helps cooling no?, more impingement?... Last edited by MadDogMe; 03-21-2003 at 03:52 AM.

03-20-2003, 08:19 PM	#5
Alchemy Cooling Savant Join Date: Oct 2002 Location: Boston Posts: 238	OTMOPO3OK is mostly right. So long as your tubing is the same diameter, the velocity of the water is going to be the same at any point in the tubing. Likewise, the pressure drop across the HC and across the WB will be the same regardless of what order they are in. So as far as heat transfer properties go, they will be the same no matter what order they are in. More important factors are how much flow restriction is caused by bends in the tubing. Avoid those. There's heat input due to the pump is often overrated, I think, except in cases where people are using huge pumps near the heat output of the processors they're trying to cool. Put the pump before the rad if you know it won't cause more problems with your piping system. Otherwise, don't worry about it. Alchemy

03-21-2003, 07:18 AM	#7
chesspatzer Cooling Neophyte Join Date: Mar 2003 Location: Planet Unreal Posts: 8	From a Dusty MechEng A Simplified Generalization: If I'm applying the flow formula correctly: m=pAV, where m is flow rate, p is 'rho' which is the density of water (ignore in this case), A is cross sectional area of the tube, and V is velocity. Flow is constant throughout the system, that is, you don't have a greater amount of water flowing out of the pump than you have flowing in to it and the same for any component. Since the flow is constant and the density is (basically) constant, the above formula can be reduced to AV=c so that the Area times the Velocity equals some constant. Therefore, as the area increases, the velocity decreases and vice versa. The point is this, the arrangement of components, Pump->HC->CPU or Pump->CPU->HC, is irrelevant on velocity. It is the cross sectional area of the Block the governs velocity. It's been a little while since I've used my engineering background but here's what I think really matters. Since you want to draw heat off of the CPU, a "convective heat transfer process", the basic formula is: Q=hA(dT), where Q is the rate of heat transfer (bigger is better), h and A are dependent on the block characteristics, and (dT) is delta T or temperature difference. The only thing that you can control by your setup is the delta T. The bigger the delta T, the better. Now, what is delta T? It is the temperature difference between the surface temperature of the CPU block and the water flowing across it. Specifically it is t(s) minus t(w) where t(s) is the temperature of the surface and t(w) is the temperature of the water. To make this number bigger, our goal, we can make the CPU hotter, not what we want, or make the water cooler! Therefore, in my old, rusty, dusty, mind - Pump->HC->CPU would be preferred because it puts the lowest possible water temperature into contact with the CPU.

03-21-2003, 07:31 AM	#8
winewood Cooling Savant Join Date: Nov 2002 Location: in my chair Posts: 574	Kudos for keeping it simple. Thank you. I believe you are one of the few that don't explode my head with characters and equations that arent confusing. (baby steps) I believe from the discussions I read before, the thing that goes down in a loop from pump to end unit is pressure. So.. flow rate stays the same, but the pressure drops as it flows away from the source. Keeping the pump -> block the psi will be highest, at this point, and that can help with the transfer of heat. The radiator is one of the larger contributors to psi drop, therefore keeping it after the block contributes to a larger psi at block. The before mentioned configuration sounds logical to me, and I use it myself. Did I totally botch that understanding? Corrections welcome. __________________ -winewood-

03-21-2003, 10:03 AM	#9
OTMOPO3OK Cooling Neophyte Join Date: May 2002 Location: Brooklyn, NY Posts: 45	thx for replies guys... chesspatzer, you too for explanation but i have to agree with winewood i still feel that putting CPU after pump gives it more turbulance ( i use nozzle)... i wish i could test both but i have very limited supply of tubing i got from my Chemistry lab professor and don't want to waste it trying out things...maybe i'll get some cheap silicone and try that... __________________ Keyboard not found. Press F1 to continue... Last edited by OTMOPO3OK; 03-21-2003 at 10:09 AM.

03-21-2003, 10:18 AM	#10
Cova Cooling Savant Join Date: May 2002 Location: Canada Posts: 247	I'm pretty sure I read on these forums somewhere by one of the other people who can never have enough math in a post (I believe it was BillA) a different forumula for calculating the heat transfer from the block to the coolant, and that pressure was involved in it. It's not as simple as a convective heat transfer - it would be convection getting the heat in the coolant that actually touches (at the molecular level) the block to the rest of the coolant flowing through, but it is conduction getting the heat from the edge-molecules of the copper into whatever coolant molecules come into contact with it. Either way, I don't think it makes a measureable difference (just like I don't think pump-heat is a measurable enough difference to argue about where it should go), and recommend that the pump is placed either for aesthetic purposes in the case of big windows, or wherever simplifies tube-routing the most for cooling purposes. As for pump-created turbulence, I don't think that will make a measurable difference either unless you pretty much just put the pump-output right on/in the block (as a few here have tried).

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)