Go Back   Pro/Forums > ProCooling Technical Discussions > General Liquid/Water Cooling Discussion
Password
Register FAQ Members List Calendar JavaChat Mark Forums Read

General Liquid/Water Cooling Discussion For discussion about Full Cooling System kits, or general cooling topics. Keep specific cooling items like pumps, radiators, etc... in their specific forums.

Reply
Thread Tools
Unread 07-10-2002, 06:09 PM   #26
airspirit
Been /.'d... have you?
 
airspirit's Avatar
 
Join Date: Jul 2002
Location: Moscow, ID
Posts: 1,986
Default

I have to agree that most naval techs had no frickin' clue. Sad, but true. I just wasn't your average naval tech. What I learned about pumps was in connection to nuclear reactors (talk about mission critical water cooling!). There are dozens of pumps required to keep one up, and in part of the training they went heavy into system design. Most people slept (or tried to ... hard to do when standing up) through alot of that. I found it interesting. I guess some of that is paying off.
airspirit is offline   Reply With Quote
Unread 07-10-2002, 07:08 PM   #27
myv65
Cooling Savant
 
Join Date: May 2002
Location: home
Posts: 365
Default

Bigben2k stated this earlier, but it seems to have been lost somewhere along the line. If you put those two pumps in series and either one dies, the flow will plummet. These pumps work by spinning an impeller inside a casing. The casing takes input at the center of the impeller and sends it out at the periphery. If impeller no spinny, flow no go (very well).

There's nothing wrong with running them in parallel, though I'd strongly recommend a check valve at each pump's outlet. Again, if one pump died it would be a "short-circuit" and you'd have some flow going where you wanted with the rest flowing backwards through the first pump and into the second pump's suction. If the resistance of the stopped pump was comparable to the circuit resistance, total flow would go up, but flow through each circuit would be less than if the dead pump wasn't present.

In parallel, total flow will be higher than either pump would do on its own provided the discharge pressure of the combined pumps was below the dead-head pressure of both. In series, the total flow would also be higher than either pump individually so long as the zero-head flow of both pumps was higher than the resulting total flow If one pump had a zero-head discharge lower than what the other pump would do under load, the "weak" pump would be nothing more than another (albeit small) resistance to the "strong" pump.
myv65 is offline   Reply With Quote
Unread 07-10-2002, 09:30 PM   #28
Skulemate
Cooling Savant
 
Skulemate's Avatar
 
Join Date: Jul 2002
Location: Toronto, Canada
Posts: 381
Default

If you're interested in reliability my opinion is that the pumps should be arranged in parallel, regardless of whether or not this is optimum. That's because it's easier to set up check valves to keep the flow going in the right direction if one pump dies than it is to set up a bypass, which would be needed if the pumps are in series (I doubt that you'd get a whole lot of flow through a dead pump, as myv65 and others have stated).

And one more thing. Myv65's statements about pumps in series versus parallel are dead on... couldn't have said it better myself.
__________________
Michael E. Robbins
M.A.Sc. Candidate, University of Toronto

12.1 GHz of AMD's finest (17.7 GHz total) crunching proudly for the AMDMB.com Killer Frogs
SETI BOINC: Dual Opteron 246s (Iwill DK8N) | XP2800+ (Shuttle SN41G2) | 3x XP2400+ (ASUS A7N266-vm)
SETI BOINC: 2x P4 2.8E (ASUS P4R800-vm) | Crunching 24/7
Skulemate is offline   Reply With Quote
Unread 07-11-2002, 04:34 AM   #29
Volenti
Cooling Savant
 
Volenti's Avatar
 
Join Date: Jan 2002
Location: in a nice cool spot
Posts: 427
Default

Quote:
(I doubt that you'd get a whole lot of flow through a dead pump, as myv65 and others have stated).
The test;

As shown in pic, the square grey pump is rated at 2000L/H, the black pump at 2800L/H. 5/8'' tubing, timing to fill a 9L bucket.

flow with both pumps going; 1620L/H.

flow with black pump "dead"; 939L/H

flow with grey pump "dead"; 1120L/H
Attached Images
File Type: jpg seriespumps.jpg (17.0 KB, 95 views)
__________________
feel free to icq/msn me, I'm always willing to toss around ideas.
Volenti is offline   Reply With Quote
Unread 07-11-2002, 08:28 AM   #30
bigben2k
Responsible for 2%
of all the posts here.
 
bigben2k's Avatar
 
Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
Default

All right Volenti!!! There's a guy that's willing to go the extra mile!

Still though, like I said earlier, a pump failure is a risk, and if you are going to go through a Risk Assesment exercise, you'll find that the pumps are very reliable, and that their failure is highly unlikely. That being said, setting up these pumps in series poses NO RISK since there will still be some flow, as Volenti just demonstrated, if one of them goes dead.
bigben2k is offline   Reply With Quote
Unread 07-11-2002, 09:43 AM   #31
airspirit
Been /.'d... have you?
 
airspirit's Avatar
 
Join Date: Jul 2002
Location: Moscow, ID
Posts: 1,986
Default

Take a look inside of your pump at the impeller. You'll find that they give next to no resistance. Conversely, this is why they give lower pressure than a volumetric pump: they are really inefficient in design, and designed so that they can be set inline to increase pressure. It is all too easy when setting them up in parallel to have one of them back-feed into the other unless they are identical pumps. If in parallel, when one goes out it'll turn into a back-feed loop for the one that works. This will cause a good deal of the pressure from the working pump to be fed directly in a loop back to itself bypassing the block/radiator/etc. (path of least resistance, just like electrical current). Even with a valve to prevent back-flow, it'll just reduce effeciency in the pumps while they both work. Series is ALWAYS better when dealing with centrifugal pumps. ALWAYS.
airspirit is offline   Reply With Quote
Unread 07-11-2002, 09:52 AM   #32
bigben2k
Responsible for 2%
of all the posts here.
 
bigben2k's Avatar
 
Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
Default

Here here!

These mag drive centrifugal pumps are not the most efficient, but the alternatives are either too expensive and/or noisy.

Just ask Cieprus...
bigben2k is offline   Reply With Quote
Unread 07-11-2002, 06:53 PM   #33
Skulemate
Cooling Savant
 
Skulemate's Avatar
 
Join Date: Jul 2002
Location: Toronto, Canada
Posts: 381
Default

Well, I must say that I am a bit surprised by those results, but they do show almost 50% flow reduction when the large pump dies and only the weak pump remains... hardly next to no resistance. I agree that a check valve meant to prevent back-feed loops in a parallel setup will reduce performance, but they're not that bad in terms of resistance. I'll will however agree with bigben's conclusion that there's really not a whole lot of risk setting the pumps up in series. And while it may be fair to say that series is always better when dealing with mag-drive centrifugal pumps, I don't think that your statement is entirely accurate as is airspirit.
__________________
Michael E. Robbins
M.A.Sc. Candidate, University of Toronto

12.1 GHz of AMD's finest (17.7 GHz total) crunching proudly for the AMDMB.com Killer Frogs
SETI BOINC: Dual Opteron 246s (Iwill DK8N) | XP2800+ (Shuttle SN41G2) | 3x XP2400+ (ASUS A7N266-vm)
SETI BOINC: 2x P4 2.8E (ASUS P4R800-vm) | Crunching 24/7
Skulemate is offline   Reply With Quote
Unread 07-11-2002, 10:32 PM   #34
gone_fishin
Cooling Savant
 
gone_fishin's Avatar
 
Join Date: May 2002
Location: Da UP
Posts: 517
Default

How would this configuration do? Would it increase the volumetric flow over that of a single pump to approach double? And why not two rads to offset extra pump heat.
Attached Images
File Type: jpg combined flow.jpg (22.9 KB, 85 views)
gone_fishin is offline   Reply With Quote
Unread 07-12-2002, 01:29 AM   #35
Volenti
Cooling Savant
 
Volenti's Avatar
 
Join Date: Jan 2002
Location: in a nice cool spot
Posts: 427
Default

Quote:
How would this configuration do? Would it increase the volumetric flow over that of a single pump to approach double? And why not two rads to offset extra pump heat.
hmm, I'm currently in the process of designing/making a P4 water block with dual 1/2'' in and out (though just using my single big pump) I can do some exprimentation then.

At a guess I'd say you should get around 1.5x the flow at least, assuming the water block has extremly low flow resistance.

I't's getting hard to design an efficient water cooling loop that doesn't cause these pumps to loose well over half their flow...
__________________
feel free to icq/msn me, I'm always willing to toss around ideas.
Volenti is offline   Reply With Quote
Unread 07-12-2002, 02:22 AM   #36
gone_fishin
Cooling Savant
 
gone_fishin's Avatar
 
Join Date: May 2002
Location: Da UP
Posts: 517
Default

Yeah, dual inlets and outlets on the block would kill the need for the Y and be the same principle. I'll be interested in some test results when you post them
gone_fishin is offline   Reply With Quote
Unread 07-12-2002, 06:51 AM   #37
Volenti
Cooling Savant
 
Volenti's Avatar
 
Join Date: Jan 2002
Location: in a nice cool spot
Posts: 427
Default

Actually I'm still stuck on some of the design concepts, namley the design of the "whatever" directly above the core that I'll have all this high volume water rushing through. I may make a simple prototype to test the overall flow side of things before I get all cought up in an "ideal" design...
__________________
feel free to icq/msn me, I'm always willing to toss around ideas.
Volenti is offline   Reply With Quote
Unread 07-12-2002, 08:21 AM   #38
bigben2k
Responsible for 2%
of all the posts here.
 
bigben2k's Avatar
 
Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
Default

Quote:
Originally posted by gone_fishin
How would this configuration do? Would it increase the volumetric flow over that of a single pump to approach double? And why not two rads to offset extra pump heat.
It certainly wouldn't be my first choice. That's like running pumps in parallel. Also, because of the rad placement, they are in parallel too, that part is good.

If you're looking to increase performance, get a pump with a bigger head: your actual flow rate will be much closer to the pump's spec at 0 head. Or, do what I'm considering: a combination of parallel and series (like 4 ViaAqua pumps).

Good pumps are not only hard to find, they're also extremely expensive. I've specd out an Iwaki, amazing pump, but $300, so no way.
bigben2k is offline   Reply With Quote
Unread 07-12-2002, 09:37 AM   #39
airspirit
Been /.'d... have you?
 
airspirit's Avatar
 
Join Date: Jul 2002
Location: Moscow, ID
Posts: 1,986
Default

http://www.puddleandpond.com/nursery_pro_pumps.htm

Try the npu500. At 3 ft head, I get crazy flow, even in a 1/2" sys w/ a 3/8" block. These stats, if anything, are UNDERstated. These pumps CAN be ran dry/inline. They are wrapped in a big rubber grommet (very nice) and make virtually no noise (some vibration, so put some packing foam beneath as a shock absorber and it is silent). Very beefy and reliable. You'd be hard pressed to find any 2 pumps of the correct size that'll outsquirt this sucker. I got mine for under $80.

Oh, as far as the design above, having the radiators in parallel is the best config for them. Rads are more efficient when there is a greater temp diff between the two fluids (air/water). That allows the warmest water to hit both. As far as the pumps, you REALLY need to set them up in series. Truly, you do.

Last edited by airspirit; 07-12-2002 at 09:49 AM.
airspirit is offline   Reply With Quote
Unread 07-12-2002, 09:51 AM   #40
airspirit
Been /.'d... have you?
 
airspirit's Avatar
 
Join Date: Jul 2002
Location: Moscow, ID
Posts: 1,986
Default

Here is a diagram. Because two pumps are used/creating heat, I placed them before the rads. The res is before the pumps to ease priming of the system. Notice the pumps are series and the rads are parallel. I think you'll get maximum results from this config.

edit: You can also put the blocks in parallel ... with this much cooling power, it might not be a bad idea.
Attached Images
File Type: jpg diag.jpg (7.8 KB, 65 views)
airspirit is offline   Reply With Quote
Unread 07-12-2002, 02:13 PM   #41
gone_fishin
Cooling Savant
 
gone_fishin's Avatar
 
Join Date: May 2002
Location: Da UP
Posts: 517
Default

Quote:
Originally posted by bigben2k

It certainly wouldn't be my first choice. That's like running pumps in parallel. Also, because of the rad placement, they are in parallel too, that part is good.

If you're looking to increase performance, get a pump with a bigger head: your actual flow rate will be much closer to the pump's spec at 0 head. Or, do what I'm considering: a combination of parallel and series (like 4 ViaAqua pumps).

Good pumps are not only hard to find, they're also extremely expensive. I've specd out an Iwaki, amazing pump, but $300, so no way.

You missed the point of the special Y adapter going from two 1/2" id to one 3/4" id. As long as the block doesn't restrict the flow in this configuration then I see no reason for it not to work.
I get plenty of flow with my pump, I am only speculating that two weaker pumps may perform better than one.
gone_fishin is offline   Reply With Quote
Unread 07-12-2002, 03:30 PM   #42
airspirit
Been /.'d... have you?
 
airspirit's Avatar
 
Join Date: Jul 2002
Location: Moscow, ID
Posts: 1,986
Default

Two weaker pumps will not work better than one good one. The combined force isn't:

F[pump1] + F[pump2] = F[total]

It's more like:

F[pump1] + (F[pump2] * %loss) = F[total]

The pumps aren't nearly as efficient if I hard load of water are hitting them. This applies in series.

In parallel, you'll find that the Y to a larger hose size WILL help with potential backflow problems, but it just isn't as efficient. The system head will NOT be increased as much as if they were in series, even considering the loss of power to the second pump (assuming pump 1 is of equal or greater strength). You'd primarily be increasing your GPH while your head remains static, assuming that the hose volume at the Y is double what it is in each parallel stream. With the pumps in series, your GPH will be equal to the GPH of the stronger pump while your head will be increased in the manner listed above. Note that the excess GPH will not help in terms of cooling. You don't want a huge water slug hitting your radiator since it has tight channels. I imagine that will become a bottleneck (unless you have a LARGE auto radiator w/ intake/outlet of 3/4" or greater).

The only reason to run 2xcentrifugal pumps over 1 pump is for redundancy in case of pump failure. You will be more efficient with one pump. The other factor that comes into this equation is that each pump releases heat into the system. With two pumps whose combined output equals one larger pump, you would get about 1.5x the heat output. Extra heat in the system does not help in what we're trying to do.

Honestly, if you have two rads and two pumps and you insist on using them all, you might want to run your CPU off of one set and your GPU/Chipset off the other. If you insist on running it as one closed system, do yourself a favor and run the pumps in series and the rads in parallel. I promise you that it is the most efficient way, and it WILL get you the best results.
airspirit is offline   Reply With Quote
Unread 07-12-2002, 03:45 PM   #43
Skulemate
Cooling Savant
 
Skulemate's Avatar
 
Join Date: Jul 2002
Location: Toronto, Canada
Posts: 381
Default

Quote:
Originally posted by airspirit
With the pumps in series, your GPH will be equal to the GPH of the stronger pump...
That's not correct. It has been said, but I will re-iterate it. For pumps in series, to figure out how much flow and head you'll get from the pumps you must add the head produced by each pump produced at a given flowrate (i.e. say from your pump curve you know that for a flow of 200GPH pump 1 produces 3' head and pump 2 produces 1'... the total head imparted on the water is 4', and the flow is of course 200GPH). Now, for pumps in parallel, you must add the flowrates produced from each pump that correspond to the total resistance of the system (i.e. if the system has a head of 4', then say pump 1 has a flow of 150GPH and pump 2 50GPH... then the total flow is 200GPH and the head is of course 4'). Now, this is not quite as simple as all that, since the pump's flow/head relationship is a curve, and so is the system flow/resistance relationship. Confused? Whatever, it doesn't matter... all I am really saying is that we do not have enough information to predict whether parallel or series is better from theory, and that is why I will continue to insist that both should be tried before any conclusions are made.

P.S. This isn't aimed at anyone in particular, but come on guys... stop it with the numbers like 1.5* this or 10* that without something to back up your claim... please...
__________________
Michael E. Robbins
M.A.Sc. Candidate, University of Toronto

12.1 GHz of AMD's finest (17.7 GHz total) crunching proudly for the AMDMB.com Killer Frogs
SETI BOINC: Dual Opteron 246s (Iwill DK8N) | XP2800+ (Shuttle SN41G2) | 3x XP2400+ (ASUS A7N266-vm)
SETI BOINC: 2x P4 2.8E (ASUS P4R800-vm) | Crunching 24/7
Skulemate is offline   Reply With Quote
Unread 07-12-2002, 03:48 PM   #44
airspirit
Been /.'d... have you?
 
airspirit's Avatar
 
Join Date: Jul 2002
Location: Moscow, ID
Posts: 1,986
Default

Eh, I oversimplified a bit, but good catch.

From experience, I assure you that with centrifugal pumps, you will get better efficiency in series. You can test it if you want, but you'll be wasting time and money.
airspirit is offline   Reply With Quote
Unread 07-12-2002, 03:59 PM   #45
Skulemate
Cooling Savant
 
Skulemate's Avatar
 
Join Date: Jul 2002
Location: Toronto, Canada
Posts: 381
Default

Ok... that may well be the case. I've never tried it, mainly because I only own one pump . Don't forget though, every system is very different...
__________________
Michael E. Robbins
M.A.Sc. Candidate, University of Toronto

12.1 GHz of AMD's finest (17.7 GHz total) crunching proudly for the AMDMB.com Killer Frogs
SETI BOINC: Dual Opteron 246s (Iwill DK8N) | XP2800+ (Shuttle SN41G2) | 3x XP2400+ (ASUS A7N266-vm)
SETI BOINC: 2x P4 2.8E (ASUS P4R800-vm) | Crunching 24/7
Skulemate is offline   Reply With Quote
Unread 07-12-2002, 04:21 PM   #46
airspirit
Been /.'d... have you?
 
airspirit's Avatar
 
Join Date: Jul 2002
Location: Moscow, ID
Posts: 1,986
Default

True, but if you drop a rock off a cliff in Colorado and another off a building in Tibet, they both fall down.

I'm just stating general facts about how these things work. It would be a rare system indeed that could prove me wrong, and I'd be VERY interested in seeing it. To make parallel pumps work better than series you would require three things: two identical pumps, a water block with a channel diameter equal to the combined flow from both pumps, and a MASSIVE radiator (think of the main radiator on a F350 or so). Even then, it would probably be a dead heat with the pumps in series with the same components. On the small scale we're dealing with, I dare say it would be impossible for a parallel pump arrangement to give better results.
airspirit is offline   Reply With Quote
Unread 07-12-2002, 07:23 PM   #47
gone_fishin
Cooling Savant
 
gone_fishin's Avatar
 
Join Date: May 2002
Location: Da UP
Posts: 517
Default

I may just try this out to see where it goes. I have a block that could easily handle it and all I would need to do is buy two identicle pumps and another heatercore. My block is such that I can have any diameter barbs of choosing with the proper top depth. This does have me extremely interested. The pump I have now is massive in terms of power but if I could equal it with two smaller pumps at the same cost I would have redundancy for better overall system protection. Also two of the heatercores that I use are less than a black ice in cost but they are much better. Strong marketing points for sure.
Any thoughts?
gone_fishin is offline   Reply With Quote
Unread 07-12-2002, 07:27 PM   #48
Skulemate
Cooling Savant
 
Skulemate's Avatar
 
Join Date: Jul 2002
Location: Toronto, Canada
Posts: 381
Default

I would love to see whether or not it'll perform. Will you be arranging it as in your drawing from a few back (although without the "y" since your block will take the two inputs right...)?
__________________
Michael E. Robbins
M.A.Sc. Candidate, University of Toronto

12.1 GHz of AMD's finest (17.7 GHz total) crunching proudly for the AMDMB.com Killer Frogs
SETI BOINC: Dual Opteron 246s (Iwill DK8N) | XP2800+ (Shuttle SN41G2) | 3x XP2400+ (ASUS A7N266-vm)
SETI BOINC: 2x P4 2.8E (ASUS P4R800-vm) | Crunching 24/7
Skulemate is offline   Reply With Quote
Unread 07-12-2002, 07:34 PM   #49
gone_fishin
Cooling Savant
 
gone_fishin's Avatar
 
Join Date: May 2002
Location: Da UP
Posts: 517
Default

Yes. And I think I will machine a custom Y also to still have a central inlet.
gone_fishin is offline   Reply With Quote
Unread 07-12-2002, 10:02 PM   #50
Volenti
Cooling Savant
 
Volenti's Avatar
 
Join Date: Jan 2002
Location: in a nice cool spot
Posts: 427
Default

Just to continue on with some measured results with my 2 pumps;

Airspirit is on the right track with his "With the pumps in series, your GPH will be equal to the GPH of the stronger pump..."

BUT, it depends on the restriction to flow, in my earlier test setup the flow with both pumps going ended up slightly lower than with just the big pump by it's self (1620L/H v 1800L/H), however, when you put both pumps to work through a water block,rad ect you end up with a net gain (900L/H large pump, 1220L/H both pumps)

I havn't done any paralell tests since my prototype block isn't done yet, but getting both pumps efficiently through it would require 3 half inch fittings in and out,since the large pump really needs 2 half inch fittings by it's self, which I suppose won't be that hard given the area that you can play with on a P4...

oh in case your wondering, I don't have to worry about the rad setup since I use bong cooling, everything just get's dumped into the top of the bong. which is one of the reasons why I'm being obsessive about flow rate (:
__________________
feel free to icq/msn me, I'm always willing to toss around ideas.
Volenti is offline   Reply With Quote
Reply


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

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -5. The time now is 05:01 PM.


Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com
If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk...
Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...