Flowrate, Pressure, PQ Curves, & You: Understanding Watercooling Pump Performance

[@ninjaduck3534]

So informative and well demonstrated- thank you!

[@jassilamba80]

Nicely done,explains pressure and flow in a very easy to understand way. Nide demo as well.

[@robin1122211]

thanks for the informative video! But i got one question: does the flow rate have any impact on your temperatures? eg a 200 LPH 1.5m head preassure pump vs a 800LPM 5m head preassure pump

[@tdomingues89]

it does, but when you pass the 1GPM = 3.78 L/M or 228L/H , the return aren´t that big, like 0.5C - 1C. martinsliquidlab.petrastech.com/MartinsFlowRateEstimator.html

[@mitchellbarnow1709]

What an incredible setup on your desk! I don’t know anything about computer cooling systems. I’m too early for Tesla videos? I’ve got a few more years to go!

[@ShaneHerald]

wow ....I have to say thanks for the video!...all the work you did and the time it took to set that up and make this video ...very informative ....thank you! I cant believe this only has 4960 views....ima have to post this on g+ get you some more views .....

[@AlexVenz]

+Shane Herald It is sad that the video hasn't gotten more views... but mostly because questions about this sort of thing are still really common in water cooling forums and social media groups. Go forth and educate!

[@ShaneHerald]

Alex Venz so true ..but actually seeing it happen is a much easier way of understanding what is actually happening ....im amazed at every thing you set up there ....that's insane! just to do this video....lol...

[@LebowskyDude]

Excellent video man. Good luck with the rest :D

[@MosquitoMade]

Very good information here and well presented. Thanks!

[@tdomingues89]

@Alex Venz I need some help, i am going to review like 12 pumps from aliexpress, and i wanted to ask, if i just need 1 pressure sensor to read the pressure drop of a cpu block. Or i can use has a refrence the pressure of the exit of the pump minus the pressure output of the cpu block at a given RPM or duty cycle?

[@AlexVenz]

You're going to need two pressure sensors.

[@tdomingues89]

@@AlexVenz For real ? XD Reeee. It took me 1 month to get my stuf, a T fitting 1/4 and a pressure sensor 1.2MPa. Because if i know the pump exit pressure at a determinat flow rate/RPM i could calculate the exit pressure on the cpu block. The worse part, is how the F i am going to calibrate my pressure sensor XD

Wow, this video is incredibly informative. Thank you for this amazing class. I'd like to know what's the effect of lowering the RPM of the pump (in PWM controlled pumps) in the curve you showed. Would that lower the the whole curve or just move the operating point to a lower flow rate?

[@AlexVenz]

Generally speaking, reducing the impeller RPM will drop the whole performance curve, which results in a lower flowrate through the system.

@@AlexVenz But the system pressure drop curve stays the same, right? So the effect of lowering RPM of the pump will be travelling down the system pressure drop curve. Am I right? The reason I'm asking this is that from my experience, there is a minimum flowrate from wich there is no noticeable temperature gain in a custom water cooler. If I can find that value, I can then optimize my system for the lowest possible noise without any performance losses.

[@AlexVenz]

@ Correct, the pressure drop curve of the components in the loop stays the same and the point of intersection between the pump's performance curve and the component pressure drop curve ends up lower on the pressure drop curve. As for flowrate vs. component temperatures, the point of diminishing returns is usually around 1 us gallon per minute for common PC water cooling parts. Anything above that 1GPM flowrate doesn't typically do much and performance tends to degrade quickly below about 0.5GPM. I'm not up on who's still doing good independent testing these days, but some of ProCooling's old waterblock performance curves can still be found online if you want to get a feel for how flowrate impacts component temperatures in very old waterblocks.

[@jeffm2787]

Flow is a quick diminishing return on temperatures.

[@stercan]

Im running 2x D5 pumps, and when I put them from 20% to 100%, temps drop. I'm considering 4x D5 pump (for lower dB at 20% and lower temps at 100%). Do you think it will work with 2x 2x140 and 2x 9x140 rads? EK GPU wb and Eisblock XPX

[@Krepadoodle]

Great video! Well done.

[@rameshgttc]

Hello Alex, Very Impressive demonstration..! by the way the flow & pressure relations as shown in this video will be same if we consider positive displacement pumps?

[@arvutihull]

Very awesome video, thanks a lot!

[@rodrigolopez9485]

hey Alex, love your set up! I am trying to build something similar myself to do some experiments. The guage pressure the system would have to endure is between 15 and 30 psi. Is your reservoir made out of acrylic or glass? I have this concern, that an acrylic reservoir might explode, anyhow your build seems solid. I am thinking about buying a PC cooling reservoir and not going through the hassle of buying all bits and pieces, but if you tell me your wasn't to time consuming I might follow your same path. Thanks!

[@AlexVenz]

The reservoir I used here is made out of a section of 4" diameter ABS pipe that has a rubber cap on the bottom, an open top (though, I have a cover that goes on but does not seal it), a baffle glued in to allow the system to bleed air more quickly, and a couple barbed fittings threaded into the pipe wall. That said, the reservoir isn't subject to any meaningful stresses due to pump operating pressure when feeding the inlet side of the pump like this... System pressure is highest at the pump's outlet, with each component thereafter incurring a pressure drop (volumetric flowrate is constant through a system like this, pressure is not). If you were to put a sealed reservoir between your pump's outlet and a very restrictive part, then you could probably pop a reservoir... but I can't think of why you'd do that. My main reason for using this reservoir setup was to provide enough fluid volume for the pump to fill the test loop without needing to cycle the pump on and off while adding more fluid (running an Iwaki RD-30 pump dry is very bad). The size of the reservoir also made it easy to install baffles to aid in the rapid bleeding of air (baffles are up high, near the return fitting, which keeps the turbulence from the return stream and air far away from the large fitting that runs to the pump's inlet).

[@rodrigolopez9485]

@@AlexVenz Hey, thanks for the detailed answer! We are thinking about cooling our PC rigs using the pressurized air we have in the lab (instead of using pumps). We want to make sure that the reservoir won't pop even if we open the air supply a bit too much :)

[@jimweist1708]

Very cool alex nice work!

[@spartanmodz4822]

nice vid sir. very informative. i like the demo.

[@leinadreign3510]

Hi there, what is that wallscroll behind you from? Looks interesting.

[@AlexVenz]

The one directly behind me is Final Fantasy concept art by Yoshitaka Amano

[@erkinkorkmaz92]

Perfect thanks dude 🙏🙏🙏

[@enigma3432]

Hi there :) I'm designing a testbench for a watercooling system. I'm not really 100% sure how to set things up, but I'm finding out new stuff all the time and this was a super informative video by the way. And I was wondering what heating device are you using to simulate the CPU?

[@AlexVenz]

+PollyB This is a rather old waterblock test setup that doesn't employ a CPU die simulator... kinda. The setup was loosely built around an iteration of Intel's TTV (Thermal Test Vehicle) specification. As such, it uses a live Intel CPU that has had its IHS milled and a specific thermocouple embedded into the center of the IHS to provide a measurement for what is defined as Tcase, in accordance with the technical diagrams provided as part of the TTV specification. The thermocouple is then connected to external data logging equipment and monitored. This approach has its advantages, but also some significant disadvantages... like either trying to obtain an instrumented CPU from Intel or finding a really good machine shop that's willing to toss an expensive CPU into a mill and mill out the copper IHS with dry air only--you can't use the usual wet lubricants/coolants.

[@nastiak5734]

way there is such a big drop pressure on the water block?

[@AlexVenz]

Well, between the jet plate and the small microchannels, the waterblock that I had hooked up to that system presented significant flow restriction---the greater the restriction, the greater the pressure drop.

[@Jaker788]

What pump was that?

[@AlexVenz]

The pump used in the demonstration is an Iwaki RD-30.

[@jordan.rushing]

Why not vacuum the system and have no pressure except only the liquid, let it boil on the heat source and turn into a gas. Then cooled off in the radiator and back into a liquid and circle back?

[@AlexVenz]

Add in a compressor and you've just invented refrigeration cooling. Sans compressor/pump, you're approximately describing a heatpipe.

[@jordan.rushing]

@@AlexVenz Yeah I probably wouldn't use a compressor because of the noise, though that does sound nice and cold. Though I'd be worried about condensation on the chip during idle. I ended up looking around and found Novec 7100 that already has a low boiling point, so no need to worry about vacuuming and worrying about losing a seal. Sounds promising in my head. But yeah now that I think about it, it's closer to a heat pipe, just now without a vacuum and a radiator & fan on the other end to bring it back to a liquid state and gravity feed it back. Maybe have the "reservoir" right there on the chip so there's always a liquid on the chip. Only thing that worries me is the radiator handling steam. I'm by no means a mech eng, so I don't know this world at all or what's available. Oh well, maybe someone will build something like this, sounds cool.

[@jordan.rushing]

Looks like what I'm describing is a thermosiphon. Fascinating

[@jasontee9539]

How bout gas pressure and its flowrate

[@VladOnEarth]

That the Tesla dude :) Nice!

[@Tofilux]

Was here