Flow rate

[Raxxath]

I'm trying to calculate my potential flow rate and I could use a little input. I'll be using an EK-XTOP Revo D5 PWM when it arrives here. I've been doing some testing in the meantime and I'm not satisfied with my temperatures. The fact that my current pump is too weak to dislodge some air bubbles from the CPU block doesn't help, but the D5 should fix that.

 

Regardless, I believe I need to add a 240 radiator to the top of my Define C, as my current 360 is getting very warm during stress testing. I should also note that I'm focused on very quiet operation, in the 500-800rpm fan range. I think this means that'll I want to run the D5 in the 60% range, but I haven't used one yet, so I don't know for certain. Anyway, using my amazing MS Paint skills, I borrowed a graph from Extremerigs and plotted a (rough) restriction curve for my current setup and a potential one with another radiator using the following components:

 

Orange (current):

-EK Supremacy EVO on a 4770k

-MSI GTX 1080 Sea Hawk EK X

-HWLabs Nemesis GTS 360

 

Red:

An additional HWLabs Nemesis GTS 240

 

Does this seem accurate? Is it adequate? They're restrictive radiators, but they're also very slim and the low speed fan performance is top notch. Maybe someone like @For Science! could share his experience since his setup is so similar.

 

[BoredErica]

Here is a point of reference that will be somewhat useful.

 

 

These are my temps with 3x560mm Nemesis GTX, Heatkiller 1080ti block, Raystorm Pro, and a little more than 3m of tubing.

 

 

My pump speed vs temp curves (keep in mind that the low pump speed curve is for 50%, not 60% in this one):

[Raxxath]

19 minutes ago, Dark_wizzie said:

Here is a point of reference that will be somewhat useful.

 

 

These are my temps with 3x560mm Nemesis GTX, Heatkiller 1080ti block, Raystorm Pro, and a little more than 3m of tubing.

 

 

My pump speed vs temp curves (keep in mind that the low pump speed curve is for 50%, not 60% in this one):

I don't think you mentioned what pump(s) you're using though. Those RPMs look very low though, couldn't be a D5.

[For Science!]

1 hour ago, Raxxath said:

-snip-

All I can quote are my settings and flowrate measured using a koolance flow meter that is now disconnected since I got my satisfactory flow rate. So I've got a similar setup with a EKWB Z270E monoblock and 1x 240 mm BlackIceGTS and 1 x 360 mm BlackIceGTS with 2x GTX 1080 Turbo blocks from bitspower (in parallel).

 

To hit the golden standard 1 gpm (or 3.8L/min), I needed to run the single D5 pump + EKWB top at 90% pump speed. This is what I run my pump at as a constant now since pump speed has little to no impact on cooling assuming you have enough of it. Interestingly, I personally find in my setup the D5 pump to be quieter at 90% than say for example at 50%, where the 50% pump speed emits more low frequency noise which is more noticeable whereas at 90% it is a very quiet whir, it is comparable to my fans running at 800 rpm and definitely quieter than my clock ticking in the background.

[Raxxath]

3 minutes ago, For Science! said:

All I can quote are my settings and flowrate measured using a koolance flow meter that is now disconnected since I got my satisfactory flow rate. So I've got a similar setup with a EKWB Z270E monoblock and 1x 240 mm BlackIceGTS and 1 x 360 mm BlackIceGTS with 2x GTX 1080 Turbo blocks from bitspower (in parallel).

 

To hit the golden standard 1 gpm (or 3.8L/min), I needed to run the single D5 pump + EKWB top at 90% pump speed. This is what I run my pump at as a constant now since pump speed has little to no impact on cooling assuming you have enough of it. Interestingly, I personally find in my setup the D5 pump to be quieter at 90% than say for example at 50%, where the 50% pump speed emits more low frequency noise which is more noticeable whereas at 90% it is a very quiet whir, it is comparable to my fans running at 800 rpm and definitely quieter than my clock ticking in the background.

Interesting. Did you notice any temperature change from dropping the pump speed? Or did you not test for that?

[For Science!]

Just now, Raxxath said:

Interesting. Did you notice any temperature change from dropping the pump speed? Or did you not test for that?

Did not test, since the pump was noticeably louder at lower speeds and that the flow rates were not as good, For me it was fairly conclusive to leave it at the higher speed.

[For Science!]

2 minutes ago, Raxxath said:

Interesting. Did you notice any temperature change from dropping the pump speed? Or did you not test for that?

Another thing that I am still toying in my mind is that the Define C does heat up alot especially with the 360 mm rad as an intake when doing heavy GPU based calculations. I am considering putting all my radiators as exhaust and putting an airflow optimized fan as intake from the back just to balance the pressure. 

[Raxxath]

1 hour ago, For Science! said:

Another thing that I am still toying in my mind is that the Define C does heat up alot especially with the 360 mm rad as an intake when doing heavy GPU based calculations. I am considering putting all my radiators as exhaust and putting an airflow optimized fan as intake from the back just to balance the pressure. 

All of your radiator fans are currently intakes, right? And I know your reservoir is in the way, but maybe an regular exhaust fan back there would help. I don't think balancing 1 intake with 5 exhausts would be easy... or noise efficient.

 

Can you compare those Noctua IPPC fans to Gentle Typhoons in terms of noise/airflow?

[For Science!]

22 minutes ago, Raxxath said:

All of your radiator fans are currently intakes, right? And I know your reservoir is in the way, but maybe an regular exhaust fan back there would help. I don't think balancing 1 intake with 5 exhausts would be easy... or noise efficient.

 

Can you compare those Noctua IPPC fans to Gentle Typhoons in terms of noise/airflow?

The top 240 is an exhaust. Given the low rpm nature of the fans and the fact that they're up against a radiator, I think it may be possible to balance most of the flow with one fan, eventually I'll try it. Under full load the case really heats up and the easiest way to address it is to take the front panel off. I think the intake air is become pretty hot and just re-heating the water and so its not very efficient as it stands.

 

Unfortunately I don't have any gentle typhoons to test so no commend with regards to noise/airflow.

[Raxxath]

2 hours ago, For Science! said:

The top 240 is an exhaust. Given the low rpm nature of the fans and the fact that they're up against a radiator, I think it may be possible to balance most of the flow with one fan, eventually I'll try it. Under full load the case really heats up and the easiest way to address it is to take the front panel off. I think the intake air is become pretty hot and just re-heating the water and so its not very efficient as it stands.

 

Unfortunately I don't have any gentle typhoons to test so no commend with regards to noise/airflow.

I think you should have both radiators as intakes getting fresh air and a rear exhaust fan to expel the warm air, if you haven't tried it already. All that positive pressure would help exhaust as well.

 

Anyway, judging by some more great data from Xtremerigs, it just seems like flow doesn't matter all that much. There's only a 1C difference between 0.5GPM and 1GPM on the Supremacy Evo.

 

 

 

You can also see how little the flow rate matters on the radiator's cooling ability from their data, especially at low speeds. So I guess I kinda answered my own question, unless anyone has experience to the contrary...

 

 

 

[BoredErica]

14 hours ago, Raxxath said:

I don't think you mentioned what pump(s) you're using though. Those RPMs look very low though, couldn't be a D5.

Revo dual pump d5.

 

I should have been more clear, but I was in a hurry. The X Axis RPM on the second chart is for fan rpm. Pump PWM % is listed in the legend. The second graph was done with fans on two out of 3 rads active. The second chart is only for GPU core temps read in software, but a 4C delta between 1 and 2 pumps at 50% is still something.

I mean, I go through these benchmarks for blocks and rads often for 1C here and there... 4C is noteworthy.

 

[Raxxath]

7 hours ago, Dark_wizzie said:

Revo dual pump d5.

 

I should have been more clear, but I was in a hurry. The X Axis RPM on the second chart is for fan rpm. Pump PWM % is listed in the legend. The second graph was done with fans on two out of 3 rads active. The second chart is only for GPU core temps read in software, but a 4C delta between 1 and 2 pumps at 50% is still something.

I mean, I go through these benchmarks for blocks and rads often for 1C here and there... 4C is noteworthy.

 

I see now. And you have just one video card, correct? Man, that's a lot of radiator. You must have a caselabs case or something. Also interesting that 100% one pump and 100% two pumps produce the same temperatures. Even more interesting is the 10C+ difference between 50% and 100% on one pump, how can it be so huge? Your setup is a bit more restrictive than my current one, and only slightly more restrictive than my proposed one.

 

The first chart still confuses me. Are those absolute temperatures or deltas? What CPU is that? What is "SF?" Also I have no idea what's going on in the lower half of the chart, aside from the clock speed.

 

Thanks for your input.

 

[BoredErica]

2 hours ago, Raxxath said:

I see now. And you have just one video card, correct? Man, that's a lot of radiator. You must have a caselabs case or something. Also interesting that 100% one pump and 100% two pumps produce the same temperatures. Even more interesting is the 10C+ difference between 50% and 100% on one pump, how can it be so huge? Your setup is a bit more restrictive than my current one, and only slightly more restrictive than my proposed one.

 

The first chart still confuses me. Are those absolute temperatures or deltas? What CPU is that? What is "SF?" Also I have no idea what's going on in the lower half of the chart, aside from the clock speed.

 

Thanks for your input.

 

For first chart:
They are deltas over ambient. So to get true temperature, add the numbers in the chart by 26.5C. I took the temp readings in a hot room. The CPU is 7600k delidded and overclocked, with overclock and voltage in the chart. SF is Stockfish, it is like a normal 100% load (not like Prime, etc).

 

Water, Low column means only fans for one radiator are on, the pumps are at 60% pwm, and the fans are at 50% pwm (these are Silent Wings 3, so that gets me to around 500- RPM). Water, Perf means fans for all 3 rads are on and at 60% pwm (700 rpm). Max means everything to max.

 

What I think my data is saying with the second graph is that flow rates matter a lot up to a certain point. If you are running a restrictive loop with low pump speed to try to make the loop silent then you're more likely to have flow rates where it matters. As you probably know, the temperature differences between flow rates will probably be a bit less with the CPU since it is hard to cool in the first place.

 

The restriction of a loop just goes up a lot anyways once flow rates go higher and higher, to the point where having two pumps probably won't increase flow rates that much to begin with.