Thoughts on water pressure & flow rate...

[bigben2k]

Originally posted by Since87
I didn't really expect my suggestions to enable you to do more useful analysis than you already do Ben.

Well actually, it's quite useful, because "adding" curves is a manual process, and your method is just as accurate as mine, but much simpler: I can do it in Excel now! I'll just have to watch out for the error margin, that's all.

Speaking of which, I believe that the difficulty at lower flow rates comes from the lack of accuracy of the measurement: if we throw figures around as x.x (ex: 1.3), then there's an assumption of an accuracy of +/- 0.1, or at least, that's the resolution used. At lower flow rates, where I stated that the dP was 0.3 psi, that would be 0.3 +/- 0.1, which is a 33% margin of error, where at 1.6 psi, it's ~ 6%.

I didn't expect the formulae to be so simple, and I thought that the progression would not follow a defined pattern, like the electrical resistance of a lightbulb filament (to bring back the electrical analogy ).

Nice tip, thanks!

[Since87]

Originally posted by Alchemy
If you have flowrate constant, flow resistances add up just like resistances in an electrical circuit. If they are parallel, summing them up gives your overall resistance, the ratio between the temperature difference between CPU and air over the heat output of the CPU. Give or take.

I give, because I'm having a hard time taking your meaing here.

For parallel electrical resistances:

1/Rp = 1/R1 + 1/R2 + 1/R3 ...

Bill's latest heatercore results indicate that the situation for fluid flow is fairly analagous.

And then the abrutpt transition from flow to temperatures...?

[redleader]

All of it eh? Then how is it that the CPU is hot? I mean, if *all* of the heat energy is being removed, I'd expect the CPU to be at ambient temperature.

When someone cannot understand the difference between heat and temperature, we should explain it simply, give them one chance to figure it out, and then ban them if they refuse. If you don't get this simple idea, meaningful conversation will be impossible.

[8-Ball]

A little harsh maybe.

You have to understand that to some people, understanding and visualising these ideas doesn't come naturally, and you can't necessarily expect everyone to understand the first time it passes through their ears.

I'm prepared to put in a bit of time to try and explain things to the best of my capabilities, provided they actually want to learn, and are themselves prepared to put in the time to read and understand what is being said.

If this isn't the case, then a lot of time ends up getting wasted.

8-ball

[bigben2k]

Yeah, we really need an article that would be a primer to waterblock design, not just "the making of <insert name here>".

PM me if you want to collaborate.

[8-Ball]

Really busy from now until june due to finals.

But after that, I'd definitely be happy to help putting some basic articles together, outlining some of the principles and ideas. I'd have time to research it properly.

8-ball

[MMZ_TimeLord]

Originally posted by Arcturius
I don't think you have to worry about blowing off your fittings with that pump; I have one, and the 4' head spec seems pretty generous...

It does have a built-in valve to tailor flow rate, but I doubt you'll run it at anything but wide-open.

Yeah... but as I said in one of my posts, I am using clear Nylon tubing and it is not on barbs everywhere. In about 4 places it's just stuck onto some Acylic tubing and the tubing is smooth. My fear is that the grip of this tubing alone is not going to be quite enough... :shrug:

And... I have not purchased the pump yet... been busy this weekend with more wedding stuff... I get hitched on May 10th...

[bigben2k]

Originally posted by MMZ_TimeLord
And... I have not purchased the pump yet... been busy this weekend with more wedding stuff... I get hitched on May 10th...

CONGRATULATIONS!!!

Here's a piece of marital advice: stay away from this place for a while

[Alchemy]

Originally posted by Since87
I give, because I'm having a hard time taking your meaing here.

For parallel electrical resistances:

1/Rp = 1/R1 + 1/R2 + 1/R3 ...

Bill's latest heatercore results indicate that the situation for fluid flow is fairly analagous.

And then the abrutpt transition from flow to temperatures...?

Not as confused as I am.

Guess I lost track and started talking about heat transfer resistances without noticing.

Sorry.

Alchemy

[Alchemy]

Originally posted by Since87
I'd still have to say that a practical voltage source is an awfully poor analogy to a centrifugal pump.

Well, dang. You just won't let me get away with that. Why so stubborn?

Originally posted by Since87
I'm an electrical engineer.

Ah.

Can we call centrifugal pumps "crappy voltage sources that no decent EE would give the time of day," then?


Obviously, I don't want to keep using an analogy that teaches people something about fluid flow while misinforming them about electricity.

Alchemy

[Blackeagle]

Pretty interesting and amazing thread over all. One question it raised for me and didn't then at some point answer.

Pump size and the heat added to the heat load of the rad.

As long as the rad can reasonably manage the added heat then improvement in flow rate from the larger pump is helpfull. And that is variable due to the size of the rad in use, and the amount of heat being added by the pump. So this would seem to me to have no set answer. I understand the part of pump selection regarding high head rate being much more important than a high flow rate.

Can some of you guys offer at least a opinion on what size of pump would be beyond the limits of being helpfull? Where are/is the limit? If you could include some pumps you feel oversized that you've seen in use it would also help. And on the other hand ones you've seen that while fairly large are good performing.

Thanks guys.

BE

[bigben2k]

As a guideline, anything higher than 3 gpm (actual flow) is getting "up there", along with the unreasonable cost.

[Blackeagle]

Ben,

I'd be very well pleased with a set up doing 3gpm. But what pumps are there that can push a real 3gpm through a system? Little Giant # ? , Iwaki # ? or any others?

A pump that can give a real 3gpm would have a much higher rate at 0 head. So I'm unsure what pump would be needed to drive a 3gpm set up.

Not as much a cost consideration here as where the line is on gain vs gone to far.

That is not to say that when I buy a pump I won't consider the cost, I will. But would really hate to discover to late that the money spent bought a pump that crossed beyond the point of gain while costing me more.

My best guess (and that is all it is, a guess) would be that a Iwaki md-30 rlzt would be the upper limit in that companys line if it isn't over the line it's self.

Just plain unsure on this part of a strong, but proper, system.:shrug: My choice of a rad (single pass 2-342) has turned out better than I hoped for. And I'm confidant I understand what I should get in way of blocks. But don't want to screw up in my pump choice.

[bigben2k]

Originally posted by Blackeagle
A pump that can give a real 3gpm would have a much higher rate at 0 head. So I'm unsure what pump would be needed to drive a 3gpm set up.

Not necessarily. There are a number of factors involved here, so I'll try to sort some out:

#1: you should really try to get a pump that's going to be energy efficient. What that means is that, to put it simply, take the pump's curve, draw a straight line between max flow and max head, and find the point on the curve that's furthest away from the straight line: that's your "point of efficiency".

#2: some pumps have a curve that is "long", like Eheims, where you get more flow with a relatively small max head. The Iwaki #70 has a tall curve, and you really have to visit that site to see the chart: it's a good read. tip, there's also an Australian site for Iwaki that has some additional info.

#3: you really need to know what your components can do, at different flow rates. We recently brought up the point (again?) that some heatercores peak in efficiency at 1.5 gpm, so if you go with two of those in parallel, you'll be able to minimize the total pressure drop.

#4: pump heat becomes a significant factor. You would do well by being able to calculate the induced heat. Note that the pump heat appears at the flow restrictions, and not within the pump.


Personally, I lucked out with my Little Giant 2-MDQ-SC, which I found for $40, but it'll need special venting. Whatever you spend, it's already beyond cost efficient: nothing can be cheaper than a good HSF, and a watercooling rig will always exceed that. All you can do is minimize all costs. You might want to look into multiple pumps in series, if you understand the limitations.


I think that in your case, you would do better by choosing it backwards: be aware of what your waterblock can do at different flow rates, and what your core will drop in pressure, then see which pump is cheapest, while achieving a reasonable temp.

[Les]

Originally posted by Blackeagle
Ben,

I'd be very well pleased with a set up doing 3gpm. But what pumps are there that can push a real 3gpm through a system? Little Giant # ? , Iwaki # ? or any others?

Using Pa x m3/sec =Calculated "Dynamic Power(?) (Watts), and using the data from
http://www.employees.org/~slf/images/md1.gif.
Dunno but think the heat imput situation is rather unclear.

Further comparison with other manufactures adds to the confusion:



EDIT. Added Eheim graphs.

[myv65]

Originally posted by Blackeagle
I'd be very well pleased with a set up doing 3gpm.

Why? Because Ben suggested anything over 3 is overkill? Look over BillA's data on waterblocks. The curve of block effectiveness starts (obviously) steep going from zero flow, but rapidly gets shallow at a point below 3 gpm. Most blocks will be quite happy with anything around a gpm or higher. The incremental gains get ever smaller with higher flow.

Why, if you are to rely on ambient air for cooling your radiator, would you work so hard to get an extra one or two degrees C from the block?

Obviously you are free to do as you wish, but I've yet to hear a good reason why people work so hard to obtain true flow rates on the order of 3 gpm.

A good block can do a lot more with one gpm and decent pressure (to generate a decent velocity directed at the core region) that an "average" block sucking down two or three gpm.

[hara]

Not to mention that with no airflow, you will not gain much on the rad by increasing the flow and that the pump will heat up your water.

[bigben2k]

All true: cooling really includes the rad solution, and that portion can give you the extra flexibility that you want, without a larger, more expensive pump.

Just stay away from BIMs!

[Cathar]

Well, after playing around with quite a few different pumps, my view on "where the line is", can be defined in a number of ways.

If we're talking specific pumps, then a 60Hz MD-15R is about where it's at, or a 50Hz MD-20RZ for those of us living in 50Hz countries.

I honestly cannot say that you will see anything that amounts to improved cooling performance with anything specced higher than these pumps. Both pumps will give you 2gpm+ for pretty much any waterblock (including the White Water).

Beyond pumps like these pump heat becomes the "enemy" for radiator based setups. Even for evaporative cooling tower setups the heat of the pump is still a major factor in causing increasingly diminutive returns.

BillA's test data is excellent, albeit slightly misleading for flow rates in real-world setups, but this is not his fault. His test setup ensures that a supply of a constant water temperature entering the block. ie. the variable of pump heat vs pump power has been removed from the equation, and the resultant affect that this would have on the radiator causing the water to get warmer than Bill's tests would indicate.

The relationship is rather complex, and it matches off pumping power vs block performance vs radiator performance. It's a triangular relationship with some cross-dependent values (ie. block and radiator performance are both affected by flow rate), flow rate and pump heat affects radiator performance, which in turn affects the waterblock performance. When we dig deeper at the differences as we change these variable, we find that we're talking about ~0.5C changes until we get extreme. ie. there seems to be a fairly broad plateau of basically the same performance given a wide variety of pumping power.

A dirty rule of thumb is that I have yet to see any pump that draws more than 40W of power give anything but worse results.

ie. the "ideal" watercooling pump is anything that will give you:

1) > 3lpm flow rates in your setup
2) > 1.5m/5' of pumping pressure head
3) < 40W of motor power draw
4) the highest flow rates given the restrictions placed by 2/3 above.

Note that there will always be edge cases where this won't hold, but in general I stand by the above recommendations.

[redleader]

#4: pump heat becomes a significant factor. You would do well by being able to calculate the induced heat. Note that the pump heat appears at the flow restrictions, and not within the pump.

I disagree. I've hooked a meter up to a 25w Eheim 1250 and found that it used 7-8w running in my system. Plus the casing was a bit warm even through it was mounted with a case fan blowing cooled intake air over it.

Pumps are rated like TVs, power supplies, or fridges. They have an actual ussage and then a max ussage two or three times as high to cover the manufactor's ass. Like my 300w power supply that says it uses 7amps at 120v

My guess is you'd need to be at 50+ w (rated, not actual) before pump heat started to be at all relevent. Even a 100w would probably be unnoticed to anyone but Bill . . . Of course a tiny rad is another issue, but most of us run heatercores anyway.

[Skulemate]

Of course the pump power requirements will be variable. The exact power requirement will depend on the flow rate, the pressure and the mechanical efficiency of the pump. Operating close to one end of the pump's performance curve will cause the pump to draw less power... it's peak consumption is typically near its peak efficiency.

[Since87]

Originally posted by redleader
I disagree. I've hooked a meter up to a 25w Eheim 1250 and found that it used 7-8w running in my system. Plus the casing was a bit warm even through it was mounted with a case fan blowing cooled intake air over it.

Umm, what kind of meter?

[redleader]

I don't know what type. My dad borrowed it from work. Basically it sits between the wall and PC and tells you the voltage, power, frequency, watt-hours, etc.

I didn't calibrate it, but I did try it with a few lightbulbs and it was within a degree or two of their rated wattage (which should itself vary), so I think its fairly accurate.

Anyway it was really interesting. I could see power vary with CPU ussage, monitor brightness, even spining up the CDROM.

Of course the pump power requirements will be variable. The exact power requirement will depend on the flow rate, the pressure and the mechanical efficiency of the pump. Operating close to one end of the pump's performance curve will cause the pump to draw less power... it's peak consumption is typically near its peak efficiency.

I know that. But keep in mind, a larger pump will operate even farther to the side, so it should use an even smaller faction of its rated wattage. Or at least i would not expect it to increase in effciency by getting increasingly backpressured.

[Skulemate]

That's right... you'd expect to see a much lower than rated power consumption when operating on the low flow - high pressure end of the curve.

[Since87]

Originally posted by redleader
I don't know what type. My dad borrowed it from work. Basically it sits between the wall and PC and tells you the voltage, power, frequency, watt-hours, etc.

Ok, a lot of people think they can measure the current their pump is drawing and the outlet voltage and they know the power consumption.

Glad to hear you used something that could measure watts.