Personal computing discussed

Moderators: renee, Starfalcon

 
whitewind
Gerbil In Training
Topic Author
Posts: 2
Joined: Fri Mar 14, 2003 10:33 pm

Water pump flow rate....

Sat Mar 22, 2003 12:22 am

I know higer flowrate means more water going through the cpu block faster, but it also means less time in the radiator.

Would there be a large difference between 100gph and 300gph, or maybe a couple of degrees??
 
JustAnEngineer
Gerbil God
Posts: 19673
Joined: Sat Jan 26, 2002 7:00 pm
Location: The Heart of Dixie

Sat Mar 22, 2003 8:27 am

That's a good question. More flow is better, but after some point the returns diminish rapidly. I would expect that the radiator is large enough to get a close temperature approach even when the water flow rate is a bit higher. Let's start by calculating the temperature rise of the water passing through the water block. Most of the heat put out by the CPU is absorbed by the water.
    .
Q = m x Cp x Temperature rise
.
m = mass flow rate (kg/s or lb/hr)
Q = rate of heat transfer (W or BTU/hr)
Cp = heat capacity (cal/g°C = BTU/lb°F)
The heat capacity of water is about 1.0 BTU/lb°F
The density of water is about 8.33 lb/gal
1 Watt = 1 J/s = 3.41 BTU/hr
.   .
m = V x 8.34 lb/gal
Putting that together, we should get: Temperature rise (°F) = 0.41 x Q (W) / Flow (gph)

If Q = 70W and Flow = 100 gal/hr, then the temperature rise is a mere 0.29°F. If your pump actually provides the rated flow rate when putting out enough pressure to overcome the flow resistance in your system, it should be fine. However, some pumps are rated at their free discharge flow, without the restrictions of tubing, water block, radiator, etc. You should look at the pump curve from the manufacturer.
 
Buub
Maximum Gerbil
Posts: 4969
Joined: Sat Nov 09, 2002 11:59 pm
Location: Seattle, WA
Contact:

Sat Mar 22, 2003 10:09 am

I know higer flowrate means more water going through the cpu block faster, but it also means less time in the radiator.


Um... physics doesn't work like that. :-)

Yes a faster flow rate means it spends less time in the radiator, but it also has less heat to emit, because it's moving faster. As long as the water is moving fast enough (i.e. above a minimum speed) it will carry and emit exactly the same amount of heat. Each cubic centimeter either carries a lot of heat slowly, or a little bit of heat fast. Either way, it's the same amount of heat in the end.

Now that's within limits. If it's moving too slowly, it can't get the heat away from the waterblock before excess heat builds up. If it's moving too fast (which is not possible with the kind of pumps we use), there will be other losses involved that alter the picture.

Now, the next issue is just how much flow you actually get. Standard pumps that are used in watercooling computers don't have a lot of head pressure (I believe that's the correct term). And watercooling systems are notoriously flow restrictive. This means you will actually get a lot less than your 100 to 300 GPH in real use. Maybe like 10 to 15 GPH in a really restrictive system.

This means two things:
1) If you have small diameter fittings and tubing, putting in a bigger pump won't really give you very much more flow rate, because you're already at the limit of what the pump can supply at the head pressure your system puts it under. Adding a big pump will add more heat, however, because it has a bigger motor. So your net effect is more heat, not less.

2) If you want higher flow rate, you might even be able to do it without upgrading the pump, just by putting in bigger diameter tubing, bigger fittings, and less restrictive water blocks and radiator. Granted if you upgrade those, a pump upgrade might also give you a little more flow than it would have in the other case.


In general, in a water cooling system, you want heat moving as fast as possible through the water block. This means low-restrictive blocks are generally better (like Swiftech blocks). See http://www.overclockers.com for some really excellent water block analysis articles (I mean more thorough than anywhere else on the web). At the same time, you want flow rate to be slower through the radiator so you can extract maximum heat. You don't want a restrictive radiator (cause that slows the whole system), but one that expands the volume as the water comes in is good.

Since given, if all else remains the same, rate of flow is inversely proportional to volume, if a radiator expands as water enters it, that means you have the same amount of water flowing through it, but it flows through slower. Slower means better heat extraction. (Note that this doesn't contradict my opening statement because now we're talking about improving the system with different radiators, as opposed to the beginning where I was talking about different rates of flow through the same radiator.)
 
JustAnEngineer
Gerbil God
Posts: 19673
Joined: Sat Jan 26, 2002 7:00 pm
Location: The Heart of Dixie

Sat Mar 22, 2003 11:07 am

Buub wrote:
As long as the water is moving fast enough (i.e. above a minimum speed) it will carry and emit exactly the same amount of heat. Each cubic centimeter either carries a lot of heat slowly, or a little bit of heat fast. Either way, it's the same amount of heat in the end.
I believe that you just proved that there is no minimum speed. Heat In = Heat Out.
If it's moving too slowly, it can't get the heat away from the waterblock before excess heat builds up
If the water flow is really low, then the heat losses to the inside of the case from the oustide of the fittings and tubing can become significant. Eventually, you will boil the water in the block.
If it's moving too fast (which is not possible with the kind of pumps we use), there will be other losses involved that alter the picture.
I am confused by these "other losses." If the water moves too fast (say 30 ft/s), it can erode your tubing and fittings. You want it to move fast enough that flow is turbulent. (Avoid Reynolds numbers below 2100. 10000 to 100000 is good.)
Now, the next issue is just how much flow you actually get. Standard pumps that are used in watercooling computers don't have a lot of head pressure (I believe that's the correct term). And watercooling systems are notoriously flow restrictive. This means you will actually get a lot less than your 100 to 300 GPH in real use. Maybe like 10 to 15 GPH in a really restrictive system.
An easy way to measure the actual flow is to run your return line into a bucket and use a stop watch.
putting in a bigger pump won't really give you very much more flow rate, because you're already at the limit of what the pump can supply at the head pressure your system puts it under.
This depends on what you mean by a "bigger" pump. For centrifugal pumps, larger diameter impellers provide more head. Thicker impellers provide more flow at the same head. Spinning the pump faster provides more head and more flow. Generally, flow through your system is probably proportional to the square root of the supplied head. Doubling the pressure supplied will increase flow by just 41%.
In general, in a water cooling system, you want heat moving as fast as possible through the water block. This means low-restrictive blocks are generally better.
The heat will flow at the rate that it is generated. Heat In = Heat Out. You really want a block that has a high overall heat transfer coefficient (low thermal resistance) and a large contact area between the cooling fluid and the block. A block that does not restrict the water flow too much is good because it allows a higher water flow rate. Since the heat source is a flat square, for a fixed inlet temperature and fixed block design, higher water flow will cause less temperature rise and therefore result in a lower CPU temperature.
you want flow rate to be slower through the radiator so you can extract maximum heat. You don't want a restrictive radiator (cause that slows the whole system)
The flow doesn't necessarily have to be slow. What you want is a radiator that internally has a fairly large surface area for the liquid to contact the radiator. At a given length, a large diameter pipe has more surface area than a small one, but you would like to keep the flow turbulent. Laminar flow does not transfer heat as well. A long run of small-diameter tubing has more area than a short run, but it is also more restrictive to flow than shorter or larger tubing. At a given supply pressure, total flow is about proportional to the square of the tubing diameter and inversely proportional to the tubing length.

It much more difficult to get heat from the radiator to the air than it is to get it from the liquid to the radiator. That's why you have lots of fins to increase the surface area on the air side and fans to force the air through. To get the water as close as possible to the ambient air temperature, you would like to have a counter-current exchanger that has the coldest air entering the radiator exchanging heat with the coolest water leaving it and the warmest air leaving the radiator exchanging heat with the hottest water entering it. Space constraints and a desire to minimize tubing length and construction complexity frequently lead to cross-flow radiator designs. A radiator designed for an automotive application is probably so large compared to the heat output of your CPU that small design features won't matter very much. A large excess of surface area will make up for other defects.
 
Buub
Maximum Gerbil
Posts: 4969
Joined: Sat Nov 09, 2002 11:59 pm
Location: Seattle, WA
Contact:

Sun Mar 23, 2003 2:06 am

As long as the water is moving fast enough (i.e. above a minimum speed) it will carry and emit exactly the same amount of heat. Each cubic centimeter either carries a lot of heat slowly, or a little bit of heat fast. Either way, it's the same amount of heat in the end.

I believe that you just proved that there is no minimum speed. Heat In = Heat Out.


No, I state what you state in the next statement:

If it's moving too slowly, it can't get the heat away from the waterblock before excess heat builds up

If the water flow is really low, then the heat losses to the inside of the case from the oustide of the fittings and tubing can become significant. Eventually, you will boil the water in the block.


If it's moving too fast (which is not possible with the kind of pumps we use), there will be other losses involved that alter the picture.

I am confused by these "other losses." If the water moves too fast (say 30 ft/s), it can erode your tubing and fittings. You want it to move fast enough that flow is turbulent. (Avoid Reynolds numbers below 2100. 10000 to 100000 is good.)


Well there you go -- you just provided a good example yourself. :-)

That's what I'm talking about. I think it would be obvious by the statement "which is not possible with the kind of pumps we use". But there are also things like 90-degree elbows that won't give significant loss in a system that's low enough flow, but will become a bigger impact the faster the water is moving. I'm just saying there are physical minimums and maximums, obviously where the rules stop working because we're working with imperfect physical mechanisms that have physical limits, but in between those end points things will always be consistent: heat in = heat out.

putting in a bigger pump won't really give you very much more flow rate, because you're already at the limit of what the pump can supply at the head pressure your system puts it under.

This depends on what you mean by a "bigger" pump. For centrifugal pumps, larger diameter impellers provide more head. Thicker impellers provide more flow at the same head. Spinning the pump faster provides more head and more flow. Generally, flow through your system is probably proportional to the square root of the supplied head. Doubling the pressure supplied will increase flow by just 41%.


Well, once again, I think you provide a good example. I understand your need to be very specific, but when you look back at what I wrote, compared to what you restated more explicitely, aren't we pretty much saying the same thing?

How "big" a pump would it take to double head pressure from say an Eheim 1048 (a pretty standard pump)? Going to the 1250 won't do it. That gives you only a roughly 50% increase in head pressure, which is going to increase flow by only about 22%. And that's a pretty big jump in pump size and power usage. You have to go all the way to the monsterous 1060 to do more than double head pressure.

That's what I mean by a "bigger" pump. I'm talking real-life stuff you can buy, here. Not a theoretical numbers-on-the-back-of-napkins pump. There are discreet sizes you can manufacture and buy these things in. And I can tell you that going to an Eheim 1250, which isn't even the biggest, was a big step. It sucks down more than twice as much power, and just barely fit in the spot I had chosen for it.

In general, in a water cooling system, you want heat moving as fast as possible through the water block. This means low-restrictive blocks are generally better.

The heat will flow at the rate that it is generated. Heat In = Heat Out. You really want a block that has a high overall heat transfer coefficient (low thermal resistance) and a large contact area between the cooling fluid and the block. A block that does not restrict the water flow too much is good because it allows a higher water flow rate. Since the heat source is a flat square, for a fixed inlet temperature and fixed block design, higher water flow will cause less temperature rise and therefore result in a lower CPU temperature.


Thanks for the detailed explanation, but didn't you just reiterate what I said? :-) I'm talking about anecdotal evidence of what works. You're talking the engineering mechanics for what should work. Fine, no problem. They seem to give the same result.

And again with the radiator. Didn't I just say in less words what you explained? I mean there's nothing wrong with the detailed explanation, but I believe I'm correct in saying that you're basically supporting most of what I wrote.
 
drobber
Gerbil
Posts: 44
Joined: Sat May 22, 2021 7:37 am

Re: Water pump flow rate....

Wed Nov 17, 2021 1:16 am

Previously, I used special liquids, for a hot tub, it seemed an ideal option - a special tool, and even with chlorine, at a price more than affordable. But now I am sure that there is nothing better than hot tub sump pumps. I would advise you to check out sump pumps tested and approved ByRossi, a hot tub specialist and designer service.

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On