Passive Water cooling

[WallyPower]

Hi Everyone,

 

This topic is meant as a sanity check.
 

In my current situation, my PC is in a room upstairs. It already runs a custom loop. Since I’ve discovered steam remote play, I’ve been using it more and more. On top of that, my SO likes the fact that I’m more downstairs.

But steam remote play has its limitations, so I’ve decided that the next PC will be built for the living room near the TV. Because space is limited, this new to be build PC will be water cooled.

 

And that is where the passive part comes into play. Last year, we renovated your house. Because of this, we now only use our in floor heating system. So I have a heat exchanger near the TV with no function.

That’s why the idea was born to use it to passively cool the new PC.

The brand is Jaga and it has the following specs:

 

The design description by Jaga:
The Low-H2O heat exchanger consists of round seamless circulation pipes of pure red copper, fins of pure aluminium and 2 brass collectors for left or right single-sided connection 1/2”. Extended air vent 1/8" and drain plug 1/2" included.

 

A drawing with cross-section:

 

Dimensions of my heat exchanger are 240cm long, 20cm high and 11.5cm deep

 

Dissipation power:

With a water temp of 35˚ C it radiates 219 Watts passive

With a water temp of 45˚ C it radiates 508 Watts passive

With a water temp of 55˚ C it radiates 757 Watts passive

With a water temp of 75˚ C it radiates 1572 Watts passive

 

This is with a room temp of 20˚ C

PC spec wise, I’m still waiting for ZEN 3 and RDNA2/Ampere so let’s say it will most likely be equivalent to a 3700X and a 2080 Super in power consumption.
 

The pump I will be using for this is an EKWB dual D5 with PWM.

So what do you guys think? 

I you need any extra info, please ask.

 

Kind regards,

 

WallyPower

[boggy77]

i think the pump would struggle to push so much water through so much pipe. you might need a custom one.

[WallyPower]

I judged by the specs that the pump would have enough grunt;

 

EK-XTOP Revo Dual D5 PWM Serial - (incl. 2x pump)

Technical specifications:
- Motor: Electronically commuted spherical motor
- Rated voltage: 12V DC
- Power consumption: 2 x 23W
- Maximum pressure head: 7m
- Maximum flow: 1500L/h
- Maximum system temperature: 60°C
- Materials: Stainless steel, PPS-GF40, EPDM O-rings, Aluminium oxide, hard coal
- Power connector:  2 x 4-Pin Molex- and 2 x 4-Pin PWM FAN connector

 

https://www.ekwb.com/shop/ek-xtop-revo-dual-d5-pwm-serial-incl-pump

 

But if you have doubts, I'll measure the pipes and make a calculation about flow restrictions.

[burh4n]

@WallyPower Check out this old LTT video
 

I don't recall if they mention the pump they used, but it looks beefier than your standard PC water pump...My guess is that it may not be worth the headache or cost, but if you're really into trying it out, it would be cool to see. Let me know if you plan to keep going, I'd be happy to try to hash out the math with you if you want/need.

[SolarNova]

Will be interesting to see how this performs. The low efficiency of the heat exchanger at low temps means u wont have much cooling power until liquid temps rise, but as the coolant temps rise so to will ur PC temps.

 

[WallyPower]

16 minutes ago, burh4n said:

@WallyPower Check out this old LTT video
 

I don't recall if they mention the pump they used, but it looks beefier than your standard PC water pump...My guess is that it may not be worth the headache or cost, but if you're really into trying it out, it would be cool to see. Let me know if you plan to keep going, I'd be happy to try to hash out the math with you if you want/need.

I've seen that one yes, and I was shouting at the execution of LTT. They used some 40 year old cast iron dinosaur with more flow restriction than the Hoover dam. The heat exchanger I'm planning on using has been designed to be used with low volumes of water with modern central heating units at low water temperatures.

6 minutes ago, SolarNova said:

Will be interesting to see how this performs. The low efficiency of the heat exchanger at low temps means u wont have much cooling power until liquid temps rise, but as the coolant temps rise so to will ur PC temps.

 

Yes, more heat means more cooling power, So i'm very curious to what Idle and load temps are going to be. Having said that, i'm still waiting for ZEN3 and RDNA2/Ampere, so this project will go into execution fase at the end of 2020 soonest

Edited May 19, 2020 by WallyPower

[WallyPower]

I tried to do the math on the flow restriction this heat exchanger gives.

 

According to my calculations, the rad should give a pressure drop of 6620 Pascal

I put the flow at 750 L/h, because no one is going to run the pump at full speed

 

 

Techpower states on their site that the average CPU block in 2017 had a flow restriction of 8135 Pascal.

 

I think i'm fine, but if anyone disagrees, please let me know.

[SavageNeo]

14 minutes ago, WallyPower said:

I tried to do the math on the flow restriction this heat exchanger gives.

 

According to my calculations, the rad should give a pressure drop of 6620 Pascal

I put the flow at 750 L/h, because no one is going to run the pump at full speed

 

 

Techpower states on their site that the average CPU block in 2017 had a flow restriction of 8135 Pascal.

 

I think i'm fine, but if anyone disagrees, please let me know.

you are fine

[Grabhanem]

My main concern would be temps-- high end graphics cards are around 300-400W, while a PBO Zen CPU is likely around 100-150. That means you'll need around 40-45 C water temps to keep your loop under control. Even with soldered IHS on Ryzen you'll likely be looking at a pretty significant temperature delta on the CPU over that-- 45C is pretty warm for a watercooling loop.

[WallyPower]

Yes temps will likely be quite high, but to my calculations also quite constant.

It will contains quite a lot of water, my guess with the rad, res, tubing and blocks, your are looking at over 3 Liters of water. So that will dampen any spikes.

The other thing is that the water temperature will most likely hit a "wall of cooling" at around 45 degrees.

 

I fitted the data from the manufacturer in a graph and got the following extra numbers:

40 degrees, dissipation is 348 Watts

45 degrees, dissipation is 475 Watts. (they state 508 watts, but this is what I get from the fit)

50 degrees, dissipation is 618 Watts.

55 degrees, dissipation is 776 Watts.

 

So you would have to do some heavy overclocking to get the water above 50 degrees.

Googling tells me that the CPU temp will probably be 20-25 degrees above water temp, so that should give me a full load temp of 70-75 degrees.

 

Should it turn out that this is not enough, I can always mount a few 120mm on top of the radiator to make it semi-active, but the goal is to try to do everything passive.

[burh4n]

21 hours ago, WallyPower said:

According to my calculations, the rad should give a pressure drop of 6620 Pascal

Not disagreeing with you because I haven't fully delved into the calculation, but I noticed your calculation is for water at 40C, which we know won't be the case all the time nor for the entire loop. It might be good to check the calculations at the limits to make sure everything checks out at the extremes (max/min temps) as well. Honestly an integral is probably needed to calculate this properly (assuming you can calculate/determine the heat transfer function), but it certainly wouldn't be quick or easy. 

I'm curious about the materials of the radiator pipe as this will greatly impact the Reynolds number. Also, what's the inner diameter of the pipe? Your earlier post states 1/2" drain plug, but my guess is that's OD (sounded like a connector spec), and then in your calculations you have an ID of 15 mm which is larger than 1/2" (12.7 mm). 

I'm interested in throwing this in a simulation if you want to share a detailed diagram of the setup. I'm thinking to check Pressure (only) thru the radiator pipes at first and then maybe exploring the thermals. Up to you, but I'd be willing to take a swing at it if you'd like.

[WallyPower]

On 5/20/2020 at 3:43 PM, burh4n said:

Not disagreeing with you because I haven't fully delved into the calculation, but I noticed your calculation is for water at 40C, which we know won't be the case all the time nor for the entire loop. It might be good to check the calculations at the limits to make sure everything checks out at the extremes (max/min temps) as well. Honestly an integral is probably needed to calculate this properly (assuming you can calculate/determine the heat transfer function), but it certainly wouldn't be quick or easy. 

I'm curious about the materials of the radiator pipe as this will greatly impact the Reynolds number. Also, what's the inner diameter of the pipe? Your earlier post states 1/2" drain plug, but my guess is that's OD (sounded like a connector spec), and then in your calculations you have an ID of 15 mm which is larger than 1/2" (12.7 mm). 

I'm interested in throwing this in a simulation if you want to share a detailed diagram of the setup. I'm thinking to check Pressure (only) thru the radiator pipes at first and then maybe exploring the thermals. Up to you, but I'd be willing to take a swing at it if you'd like.

First of all thanks for your offer.

But I did some more thinking on this and I think that I'm going to abandon the passive idea.

I'm not abandoning the idea of using this radiator, I'm just gonna strap some fans to it.

Currently it is mounted flat to the wall, but if I put 1 or 2 washers behind it, it becomes wide enough to fit a 120mm fan on top of the heat exchanger inside the shroud.

This will turn it into the equivalent of a 2160 rad, should EKWB or Alphacool ever decide to make a radiator that long...

The fin density is only 5 FPI, so no high SP fans needed.

 

Sorry guys..

[burh4n]

@WallyPower please share your progress, I for one would be interested in seeing how things work out. Best of luck!

 

[Sophia_Borjia]

Have fan control by temperature, you should be able to run passive much of time with fan only coming on under load for long periods.  It should take a long time to reach thermal equilibrium.  Spikes of 500w won't need more cooling if few minutes long and average is 200-300.  Just a little air flow should boost heat dissipation drastically.  Fans may only be needed for some games when playing hours at sitting, some may be fine entirely passive if not putting large load on hardware continuously.

[WallyPower]

@burh4n Of course I shall share my progress, but this will be a project on the long term, so please be patient.

 

@Sophia_Borjia Haven't really thought about fan control. My initial plan was to just hook up a bunch of Artic P12 or F12 fans on the PSU and use some sort of an ultra low noise adapter. Because I'm gonna slap 18 fans on that heat exchanger, driving them all would be my first concern. It's unlikely that any motherboard will take that many fans just of the headers.  

[SavageNeo]

On 5/19/2020 at 9:43 PM, Grabhanem said:

My main concern would be temps-- high end graphics cards are around 300-400W, while a PBO Zen CPU is likely around 100-150. That means you'll need around 40-45 C water temps to keep your loop under control. Even with soldered IHS on Ryzen you'll likely be looking at a pretty significant temperature delta on the CPU over that-- 45C is pretty warm for a watercooling loop.

even 2080 ti uses much under 300w.

titan rtx uses close to 300w

rtx 8000 uses about 300w. so even if you have 2080 ti it uses much under 300-400w power. around 250

 

[Sophia_Borjia]

I wasn't talking motherboard control as much as simple temp.  There are switches that can be threaded in a tee fitting that would turn fans on at given temp.  Automotive fan control for after market fans is sometimes a probe placed between fins and a nob to set fan on temp.  Something like that.  Something like Arduino wound be perfect to control fans with relays for load, but that requires electronics and coding ability and a bit of time.  Simple way could be to hook relay for fans to motherboard fan header,  set on/off at desired temp in bios, motherboard only has to supply current to flip relay, could even control 120v powered fans that way.

[Yebi]

21 hours ago, SavageNeo said:

even 2080 ti uses much under 300w.

titan rtx uses close to 300w

rtx 8000 uses about 300w. so even if you have 2080 ti it uses much under 300-400w power. around 250

Depends on the model, my 2080Ti goes up to 330W