Running multiple pumps in series at different RPM.

[MedievalMatt]

As the title would indicate.  What are the potential issues/drawbacks with running 2 or more pumps in series but at differing RPMs.  Assuming the performance of each pump is identical (which i know its not, manufacturing tolerances and whatnot, but i digress).

As an example, take one of the D5s running at 100% speed and the second at 50%.  Would this lead to simply a 50% boost in pressure vs 1 pump, instead of running both at 100% and getting double?  Or do you run into issues with one pump overspeeding the other and creating problems.

As my system is fairly restrictive im only getting 200l/h flow rate at the highest point with my single pump at 100%.  So im hoping that doubling up will boost things, but i dont know that i need too much more flow than that.  For what its worth i have 2 420s, 2 360s, and a 120 mm rads with a GPU and CPU block.  Ideally i would want to setup a curve for the second pump based on the temps of the loop and so pump number 2 would always run slower.

Am i overthinking this?

[jaslion]

The question is here: are they running in series or parrallel?

 

If prallel aka one does on loop the other another then its whatever.

 

If in series then yes one running slower than the other will negatively impact the whole loop. They HAVE to run at the same speed always. Otherwise one isnt doing anything really except being in the way.

 

If series but one is far further down the line in the loop a lower rpm can be fine however the best setup would simply be to have the 2 pumps in series at the same speed basically next to each other and going full on.

 

 

But also do you even need that? 200l/h for a loop is BEYOND plenty good.

 

 

[Agall]

1 hour ago, MedievalMatt said:

As the title would indicate.  What are the potential issues/drawbacks with running 2 or more pumps in series but at differing RPMs.  Assuming the performance of each pump is identical (which i know its not, manufacturing tolerances and whatnot, but i digress).

As an example, take one of the D5s running at 100% speed and the second at 50%.  Would this lead to simply a 50% boost in pressure vs 1 pump, instead of running both at 100% and getting double?  Or do you run into issues with one pump overspeeding the other and creating problems.

As my system is fairly restrictive im only getting 200l/h flow rate at the highest point with my single pump at 100%.  So im hoping that doubling up will boost things, but i dont know that i need too much more flow than that.  For what its worth i have 2 420s, 2 360s, and a 120 mm rads with a GPU and CPU block.  Ideally i would want to setup a curve for the second pump based on the temps of the loop and so pump number 2 would always run slower.

Am i overthinking this?

Pumps in series increases pressure, pumps in parallel increase mass flow rate, being the general rule. Its similar to electrical power sources with voltage and current, which are really just pressure and flow rate if you boil it down... 

 

The pump before the pump in a series configuration is generally referred to as a priming (been a few years) booster pump, which you'd usually use if there was insufficient suction pressure for the main pump. Usually, the booster pump is drastically smaller than the main pump since its the main pump for a reason, but just needs a little help.

 

I don't see a scenario where you'd want to do this unless you're doing something exotic, where you'd be better off having them in parallel running the same speed as a measure of redundancy. I wouldn't see a normal PC watercooling pump having insufficient head pressure for a PC, so increasing the pressure doesn't make sense to me, but having a higher flow rate is at least a potential performance improvement, assuming there's enough temperature delta across the loop that can put extra mass flow rate to good use without just adding excessive energy to the fluid.

[NorKris]

1 hour ago, jaslion said:

 

If in series then yes one running slower than the other will negatively impact the whole loop. They HAVE to run at the same speed always. Otherwise one isnt doing anything really except being in the way.

 

If series but one is far further down the line in the loop a lower rpm can be fine however the best setup would simply be to have the 2 pumps in series at the same speed basically next to each other and going full on.

 

im running 2 in series, not at the same place in my loop, diff speeds.  no problems 

[MedievalMatt]

1 hour ago, jaslion said:

The question is here: are they running in series or parrallel?

 

If prallel aka one does on loop the other another then its whatever.

 

If in series then yes one running slower than the other will negatively impact the whole loop. They HAVE to run at the same speed always. Otherwise one isnt doing anything really except being in the way.

 

If series but one is far further down the line in the loop a lower rpm can be fine however the best setup would simply be to have the 2 pumps in series at the same speed basically next to each other and going full on.

 

 

But also do you even need that? 200l/h for a loop is BEYOND plenty good.

 

 

 

I’ve heard that more than 1g/m (200(ish) l/h) is a baseline for flow rate.

 

and the block im looking at that would hold both pumps is serial, so the flow is from one pump into the other so pressure is combined.

 

I want to run differing speeds as the one pump has 2 really horrible resonance points (25% and 70%) and runs smooth as silk at 100%.  The other is loud AF above 80%.  So sound wise they are not at all compatible.

 

typically temps are 30C over ambient so that’s not an issue.

[MedievalMatt]

54 minutes ago, Agall said:

Pumps in series increases pressure, pumps in parallel increase mass flow rate, being the general rule. Its similar to electrical power sources with voltage and current, which are really just pressure and flow rate if you boil it down... 

 

The pump before the pump in a series configuration is generally referred to as a priming (been a few years) booster pump, which you'd usually use if there was insufficient suction pressure for the main pump. Usually, the booster pump is drastically smaller than the main pump since its the main pump for a reason, but just needs a little help.

 

I don't see a scenario where you'd want to do this unless you're doing something exotic, where you'd be better off having them in parallel running the same speed as a measure of redundancy. I wouldn't see a normal PC watercooling pump having insufficient head pressure for a PC, so increasing the pressure doesn't make sense to me, but having a higher flow rate is at least a potential performance improvement, assuming there's enough temperature delta across the loop that can put extra mass flow rate to good use without just adding excessive energy to the fluid.

My thinking behind the pressure is to overcome the restriction.  So higher pressure would lead to (slightly) higher flows because fluid dynamics.  Maybe that’s incorrect and the restriction is less important than I’m intuiting.

[Agall]

1 minute ago, MedievalMatt said:

My thinking behind the pressure is to overcome the restriction.  So higher pressure would lead to (slightly) higher flows because fluid dynamics.  Maybe that’s incorrect and the restriction is less important than I’m intuiting.

I believe in one of LTT's recent videos, they talked about having a pump before and after your most restrictive block. If there's a marked drop in pressure in a certain spot, you'd be better off having those pumps spread out in the loop rather than in a proper 'series' configuration on at least how most people would think of it. 

 

The fluid dynamics of a computer watercooling loop are quite simple in comparison to others, I imagine because its low pressure and (somewhat) open. Whenever you start the pump, it shouldn't draw a vacuum, but probably lowers the water pressure in the reservoir or whatever point is the suction side of the pump.

 

They're really just there to create a differential pressure in a spot to drive mass flow rate, so unless you're getting substantial turbulence in a specific spot, the flow rate across the whole loop should be the same, so its either sufficient flow or not.

[thrasher_565]

7 minutes ago, MedievalMatt said:

 

I’ve heard that more than 1g/m (200(ish) l/h) is a baseline for flow rate.

 

and the block im looking at that would hold both pumps is serial, so the flow is from one pump into the other so pressure is combined.

 

I want to run differing speeds as the one pump has 2 really horrible resonance points (25% and 70%) and runs smooth as silk at 100%.  The other is loud AF above 80%.  So sound wise they are not at all compatible.

 

typically temps are 30C over ambient so that’s not an issue.

1gm is fine because the cpu and gpu jet plat is a bottle neck. ya every 1 gm you get is about 1c drop. unless you have a big loop or under the 1gm there no real reason to run 2 pumps.

 

like others said you have to find what speed both pumps are the same. too much presher can cause tubes to pop off and blacks to crack. not only that it eats the coating off the blocks.

[MedievalMatt]

5 minutes ago, Agall said:

I believe in one of LTT's recent videos, they talked about having a pump before and after your most restrictive block. If there's a marked drop in pressure in a certain spot, you'd be better off having those pumps spread out in the loop rather than in a proper 'series' configuration on at least how most people would think of it. 

 

The fluid dynamics of a computer watercooling loop are quite simple in comparison to others, I imagine because its low pressure and (somewhat) open. Whenever you start the pump, it shouldn't draw a vacuum, but probably lowers the water pressure in the reservoir or whatever point is the suction side of the pump.

 

They're really just there to create a differential pressure in a spot to drive mass flow rate, so unless you're getting substantial turbulence in a specific spot, the flow rate across the whole loop should be the same, so its either sufficient flow or not.

From what people here are saying it sounds like if I wanted to run 2 pumps just having segregated cpu/gpu loops and running them separately would be better than trying to make a serial config work.

 

7 minutes ago, thrasher_565 said:

1gm is fine because the cpu and gpu jet plat is a bottle neck. ya every 1 gm you get is about 1c drop. unless you have a big loop or under the 1gm there no real reason to run 2 pumps.

 

like others said you have to find what speed both pumps are the same. too much presher can cause tubes to pop off and blacks to crack. not only that it eats the coating off the blocks.

Does it matter that I have a leak shield and so the system is pretty much all of the time under a 450mbar vacuum?

[thrasher_565]

2 minutes ago, MedievalMatt said:

From what people here are saying it sounds like if I wanted to run 2 pumps just having segregated cpu/gpu loops and running them separately would be better than trying to make a serial config work.

 

Does it matter that I have a leak shield and so the system is pretty much all of the time under a 450mbar vacuum?

ya 2 loops could be better imo at lest try it and see

 

ya no idea how the leak shield works.

[ToboRobot]

"Pumps running in series will increase head (H), but not flow (Q)."

 

https://www.grundfos.com/ca/learn/research-and-insights/pumps-in-series

 

[MedievalMatt]

18 minutes ago, thrasher_565 said:

ya 2 loops could be better imo at lest try it and see

 

ya no idea how the leak shield works.

Leak shield sits on the res and pulls a vacuum.  It depends on pump speed but mine wants to be at 450milibar negative pressure.

 

So effectively things would be trying to get into the loop (like air) instead of coming out, which would be bad.

 

im gonna be getting a new case soon which will have more room to run dual loops.

 

my enthoo 719 is a bit cramped

[MedievalMatt]

18 minutes ago, ToboRobot said:

"Pumps running in series will increase head (H), but not flow (Q)."

 

https://www.grundfos.com/ca/learn/research-and-insights/pumps-in-series

 

Isn’t one kind of related to the other.  Like when you have a restriction.  That affects both pressure and flow.  So overcoming a lower flow due to a restriction you can use a higher pressure?  That was my thought process anyways.

[ToboRobot]

16 minutes ago, MedievalMatt said:

Isn’t one kind of related to the other.  Like when you have a restriction.  That affects both pressure and flow.  So overcoming a lower flow due to a restriction you can use a higher pressure?  That was my thought process anyways.

https://theengineeringmindset.com/pump-chart-basics-explained/

 

 

You may enjoy reading this.

[AnonymousGuy]

On 8/10/2023 at 4:57 AM, jaslion said:

 

If in series then yes one running slower than the other will negatively impact the whole loop. They HAVE to run at the same speed always. Otherwise one isnt doing anything really except being in the way.

This is wrong.

 

Quote

I purposely chose to use two different pumps for a couple of reasons.  For starters, there is a “forum myth”, that you need to match two like pumps when run in the same loop.  I believe this grows from an assumption that the impeller speeds need to match and also because of the fear that one stronger pump would “Push” the weaker pump beyond it’s abilities.  Fortunately, this myth is wrong.

 

Anyways @OP you don't need 2 pumps.  I run at 150lph and the biggest delta across the GPU is like 2 degrees.   Radiators don't add restriction...only cpu and gpu blocks in series and unless you're doing some wild shit like multi-gpu rendering rig isn't going to be more than a D5 can handle by itself.

[MedievalMatt]

On 8/12/2023 at 1:53 AM, AnonymousGuy said:

This is wrong.

 

 

Anyways @OP you don't need 2 pumps.  I run at 150lph and the biggest delta across the GPU is like 2 degrees.   Radiators don't add restriction...only cpu and gpu blocks in series and unless you're doing some wild shit like multi-gpu rendering rig isn't going to be more than a D5 can handle by itself.

Such a wide delta on what answer you will get RE: flow rate.  Some people are saying 300+ is needed, others (including myself) have found that overall loop performance doesnt change with lower flow.  In my case i can run as low as 80lph and the only difference i see between that and 200-220 is that the overall temp of the water in the loop is more consistent.  That is to say the loop has no "hot spot" where the rads in one area are significantly more saturated than other areas with higher flow.

Im not sure, based on my testing and the input of everyone so far in this thread, that flow rate matters a great deal.  Yes, high flow rates help with bleeding and is the only reason why i run at 200+lph now.  With leakshield i end up with trapped air in my rads which have less than ideal fitting placement (this is normal with running a strong vacuum and not a concern).  So higher flow is beneficial for me because the trapped air, while not affecting performance, makes the system louder (babbling brook noises).

I can post a detailed breakdown of my loop and things if that would be helpful.  But it sounds like as long as flow is happening the only benefit to a second pump is redundancy, and my system doesnt have an uptime requirement that would make that a necessity.

[thrasher_565]

21 minutes ago, MedievalMatt said:

Such a wide delta on what answer you will get RE: flow rate.  Some people are saying 300+ is needed, others (including myself) have found that overall loop performance doesnt change with lower flow.  In my case i can run as low as 80lph and the only difference i see between that and 200-220 is that the overall temp of the water in the loop is more consistent.  That is to say the loop has no "hot spot" where the rads in one area are significantly more saturated than other areas with higher flow.

Im not sure, based on my testing and the input of everyone so far in this thread, that flow rate matters a great deal.  Yes, high flow rates help with bleeding and is the only reason why i run at 200+lph now.  With leakshield i end up with trapped air in my rads which have less than ideal fitting placement (this is normal with running a strong vacuum and not a concern).  So higher flow is beneficial for me because the trapped air, while not affecting performance, makes the system louder (babbling brook noises).

I can post a detailed breakdown of my loop and things if that would be helpful.  But it sounds like as long as flow is happening the only benefit to a second pump is redundancy, and my system doesnt have an uptime requirement that would make that a necessity.

water cooling is a nesh thing. with loops starting at like $500+ its never recommended. the people that have done test stand buy what they said about flow. like debouer gn, j2c. but aqua computer is also a nesh thing. it cost alot, hard to find any info/videos, and the qc is questionable..

 

i have alot of aqua computer stuff and was going to get the leakshield but didn't. so i dont see much testing in water cooling anymore not like it was when i had my x58. the shift from profromace to looks largly it why no one cares.

 

over $700 in res and pump and the threads cant even match up...

other things like scratches on pannals, a use of g1/8 fittings on the filter , having to buy an upgrade cooling for your $350 controller...

fittings too close on the filter

 

there was a time when flow was king but got killed off by jet plates making a bottle neck but gave better temps and slower speeds. the raystorm cpu block was the fist and everyone just copied it.

 

more flow would give lower temps but with jetplates it limits it. and increasing the psi can cause problems like craking and erosion of plating.

 

 

[MedievalMatt]

48 minutes ago, thrasher_565 said:

water cooling is a nesh thing. with loops starting at like $500+ its never recommended. the people that have done test stand buy what they said about flow. like debouer gn, j2c. but aqua computer is also a nesh thing. it cost alot, hard to find any info/videos, and the qc is questionable..

 

i have alot of aqua computer stuff and was going to get the leakshield but didn't. so i dont see much testing in water cooling anymore not like it was when i had my x58. the shift from profromace to looks largly it why no one cares.

 

over $700 in res and pump and the threads cant even match up...

other things like scratches on pannals, a use of g1/8 fittings on the filter , having to buy an upgrade cooling for your $350 controller...

fittings too close on the filter

 

there was a time when flow was king but got killed off by jet plates making a bottle neck but gave better temps and slower speeds. the raystorm cpu block was the fist and everyone just copied it.

 

more flow would give lower temps but with jetplates it limits it. and increasing the psi can cause problems like craking and erosion of plating.

 

 

 

I think largely watercooling fell out of favor for a while.  You dont "need" to watercool a 200W GPU and a 65W CPU.  But i think we will see a resurgence with 300+W CPUs and GPU pulling down 500+W.

 

Edit - I will stand by the leakshield.  I cracked a rad once and didnt notice for a few days as the leakshield was able to compensate for the hole.  I also love it for leak testing, and being able to go on holiday and not need to prep my system or worry about it in any way is excellent peace of mind.  It also makes filling a breeze.

[thrasher_565]

1 minute ago, MedievalMatt said:

I think largely watercooling fell out of favor for a while.  You dont "need" to watercool a 200W GPU and a 65W CPU.  But i think we will see a resurgence with 300+W CPUs and GPU pulling down 500+W.

ya i agree but cost is a limit. when i got in to wc it was $150 + $100 gpu block.

but a big problem is the diy and his and now cpu socket retention.  even deliding a cpu and re applying past can make more difference then any water cooling...

[AnonymousGuy]

On 8/14/2023 at 12:36 PM, MedievalMatt said:

Such a wide delta on what answer you will get RE: flow rate.  Some people are saying 300+ is needed, others (including myself) have found that overall loop performance doesnt change with lower flow.  In my case i can run as low as 80lph and the only difference i see between that and 200-220 is that the overall temp of the water in the loop is more consistent.  That is to say the loop has no "hot spot" where the rads in one area are significantly more saturated than other areas with higher flow.

Im not sure, based on my testing and the input of everyone so far in this thread, that flow rate matters a great deal.  Yes, high flow rates help with bleeding and is the only reason why i run at 200+lph now.  With leakshield i end up with trapped air in my rads which have less than ideal fitting placement (this is normal with running a strong vacuum and not a concern).  So higher flow is beneficial for me because the trapped air, while not affecting performance, makes the system louder (babbling brook noises).

I can post a detailed breakdown of my loop and things if that would be helpful.  But it sounds like as long as flow is happening the only benefit to a second pump is redundancy, and my system doesnt have an uptime requirement that would make that a necessity.

Correct on everything.  There's no benefit really to huge flowrates.  For bleeding I'll use vise grips and pinch tubes off to create pressure waves throughout the loop and that tends to do a good job breaking up air bubbles.

 

I've had pumps fail but only if they've been run dry at some point.  And even then it's not a nuclear reactor where you *need* pump redundancy or there's a literal meltdown.  If waterflow stops...components turn themselves off when they get hot.

 

Bonus: I gave up on the Leakshield.  It really doesn't seem to work well with my PMP-600 (I'm using 1 pump to drive 3 systems in parallel...so when I say I get 145lph on my desktop it's really 450lph total coming out of the pump) and at max vacuum 450 it would bitch all-the-time about how there's a leak because the pump is running frequently blah blah.  I really just think their valves aren't meant to hold that much vacuum.  Or maybe I have so much tubing and components (3 systems...) that there's just natural losses even without there being a leak.

[MedievalMatt]

57 minutes ago, AnonymousGuy said:

Correct on everything.  There's no benefit really to huge flowrates.  For bleeding I'll use vise grips and pinch tubes off to create pressure waves throughout the loop and that tends to do a good job breaking up air bubbles.

 

I've had pumps fail but only if they've been run dry at some point.  And even then it's not a nuclear reactor where you *need* pump redundancy or there's a literal meltdown.  If waterflow stops...components turn themselves off when they get hot.

 

Bonus: I gave up on the Leakshield.  It really doesn't seem to work well with my PMP-600 (I'm using 1 pump to drive 3 systems in parallel...so when I say I get 145lph on my desktop it's really 450lph total coming out of the pump) and at max vacuum 450 it would bitch all-the-time about how there's a leak because the pump is running frequently blah blah.  I really just think their valves aren't meant to hold that much vacuum.  Or maybe I have so much tubing and components (3 systems...) that there's just natural losses even without there being a leak.

Leakshield has a maximum pump headpressure that’s around 500 millibar.  Basically they tuned it for a single d5 running at 100% and the a bit more.

 

It’s also better to be used with thick walled soft tubing or hard tubing.  They’ve also added the ability to set your own custom values for things.  It’s come a long way in the year or so I’ve had it.