Dual loop vs. 2 pumps in series vs. 2 pumps in parallel

[toms0033]

I'm on my fourth custom water cooling venture and for the first time I have two matched D5 pump / res combos to work with. One from a recently disassembled build, and one new. I really only need one, but like the look of two pump/res combos next to each other. What would be the better way to integrate them into a single system? I'm trying to figure out if it would be better to have a single loop with the pumps running in series, in parallel, or a dual CPU/GPU loop.

 

[firefoxx04]

I am taking this knowledge from the automotive world, it should work. When running two pumps (even dissimilar pumps) in series, the flow increase is almost linear. I know that is hard to believe.

This means it is almost ALWAYS better to run them in series because if one dies, the other is still running. This means all of your components will still have a chance. With two separate loops, one pump failing means that the entire loop is down until you fix it.

 

[rubix_1011]

I ran a single D5 for over 6 years and it still works fine. I think most people underestimate pumps 'dying'. Besides, your PC will shut down when temps reach thermal threshold anyway...it isn't going to cause anything to burn up.

Either 2 pumps in serial or parallel is fine, it really depends on what you want to accomplish and what your future plans are. Is there a plan to expand the loop at some point in the future? The only issue with a dual loop is that you have to now account for 2 sets of loop hardware, but if you have the room, all the power to you. Watercooling is pretty forgiving in how you want to setup your loop other than if you have excessive flow restriction or no flow at all...those are definitely bad.

 

[toms0033]

Thanks for the help. I have eliminated the a dual loop option. Now I just need to figure out if a parallel pump loop would make sense over a series. Any thoughts?

 

[rubix_1011]

If it is a single loop, a parallel pump loop is still a serial pump loop to some degree - you are still ultimately bound by the single most restrictive point in the loop.

 

[toms0033]

Please correct me if this is not correct, but a series loop would product more head pressure at the same flow rate / a parallel loop would produce a higher flow rate but the same amount of head pressure. correct?

 

[4Ryan6]

I'm on my fourth custom water cooling venture and for the first time I have two matched D5 pump / res combos to work with. One from a recently disassembled build, and one new. I really only need one, but like the look of two pump/res combos next to each other. What would be the better way to integrate them into a single system? I'm trying to figure out if it would be better to have a single loop with the pumps running in series, in parallel, or a dual CPU/GPU loop.

Have you considered keeping the 2nd for a backup, I had a D5 fail last year, it had been in service over 4 years but, thankfully I had already bought 2 D5s because I run dual loops for just that purpose, and was quickly back in business.

But in your case you would already have a backup on hand, just something to consider.

 

[rubix_1011]

Please correct me if this is not correct, but a series loop would product more head pressure at the same flow rate / a parallel loop would produce a higher flow rate but the same amount of head pressure. correct?

Yes, but a single loop is still overall, a series. You might have splits to feed the parallel sections of the loop, but ultimately, those splits re-merge back into a single, serial section of the loop.

A parallel split will produce lower restriction and depending on where it splits, you might yield faster flows for those sections if your pumps exist after the split rather than before, but you also have to realize that you're attempting to feed 2x D5 pumps that are each 1/2"ID inlet with a single 1/2"ID tube, so you have to effectively halve your feeding flow capability, which is still a continuation of the serial portion of your loop which will be more restrictive by nature, since it's serial. While

 

[toms0033]

I'm on my fourth custom water cooling venture and for the first time I have two matched D5 pump / res combos to work with. One from a recently disassembled build, and one new. I really only need one, but like the look of two pump/res combos next to each other. What would be the better way to integrate them into a single system? I'm trying to figure out if it would be better to have a single loop with the pumps running in series, in parallel, or a dual CPU/GPU loop.

Have you considered keeping the 2nd for a backup, I had a D5 fail last year, it had been in service over 4 years but, thankfully I had already bought 2 D5s because I run dual loops for just that purpose, and was quickly back in business.

But in your case you would already have a backup on hand, just something to consider.

Thanks Ryan, that's something I've been considering as well. Honestly, the flow rate and pressure from one D5 is going to be more than sufficient for this loop, but having two reservoirs side by side looks pretty cool.

Maybe it'd be better run two standalone reservoirs to achieve the look I'm after. Then mount a single D5 in a discrete location. Purchasing two standalone reservoirs and a single pump top wouldn't be that expensive of a change, and I'd have spare parts for replacement, or a future build.

On the other hand, having the redundancy of two pumps either in series or parallel would be nice and would be achievable with the hardware I already have on hand.

If I were to run two pump / res combos what in your opinion makes more sense series or parallel?

 

[toolmaker_03]

you can see for yourself how that works I have done all the configurations that are being considered here.

http://www.tomshardware.com/answers/id-2767231/serial-loop-parallel-loop.html
http://www.tomshardware.com/forum/274855-29-experimental-radiator-build

my final build or configuration, that worked best. is what I call a full parallel loop, the components are paralleled together and the radiators are paralleled together, but as rubix has so many time pointed out, the loop is a serial loop from the components to the radiators.

 

[toms0033]

Please correct me if this is not correct, but a series loop would product more head pressure at the same flow rate / a parallel loop would produce a higher flow rate but the same amount of head pressure. correct?

Yes, but a single loop is still overall, a series. You might have splits to feed the parallel sections of the loop, but ultimately, those splits re-merge back into a single, serial section of the loop.

A parallel split will produce lower restriction and depending on where it splits, you might yield faster flows for those sections if your pumps exist after the split rather than before, but you also have to realize that you're attempting to feed 2x D5 pumps that are each 1/2"ID inlet with a single 1/2"ID tube, so you have to effectively halve your feeding flow capability, which is still a continuation of the serial portion of your loop which will be more restrictive by nature, since it's serial. While

That makes sense, and probably explains why I've yet to run across a commercially available parallel pump top, but just about every company offers a serial pump top.

I think I'm coming to the conclusion that if I end up running a dual pump configuration, it is going to be with the two pumps in series. However if someone has a good case for running the pumps in parallel I'd still be interested to hear what they have to say.

 

[toms0033]

you can see for yourself how that works I have done all the configurations that are being considered here.

http://www.tomshardware.com/answers/id-2767231/serial-loop-parallel-loop.html
http://www.tomshardware.com/forum/274855-29-experimental-radiator-build

my final build or configuration, that worked best. is what I call a full parallel loop, the components are paralleled together and the radiators are paralleled together, but as rubix has so many time pointed out, the loop is a serial loop from the components to the radiators.

I'm going to look over this in detail. There's a ton of info there, it's going to take a bit to read through it. Thanks for the links and hard work you did to answer this question Toolmaker

 

[toms0033]

I just noticed something interesting... everyone who's been helping to answer my question also posted in Toolmaker's thread, this must be a pretty tight corner of Tom's Hardware I've found myself in. That's pretty cool, and has given me some perspective while reading those linked threads. Thanks everyone for the help.

 

[rubix_1011]

I only point it out because we often get so hung up on whether a loop is either serial or parallel, but you really can't have a true, parallel loop because at some point, it is still a serial configuration and you're still ultimately bound by the lowest common denominator of the loop, which is the flow and restriction of that section.

 

[toms0033]

I only point it out because we often get so hung up on whether a loop is either serial or parallel, but you really can't have a true, parallel loop because at some point, it is still a serial configuration and you're still ultimately bound by the lowest common denominator of the loop, which is the flow and restriction of that section.

Thanks Rubix - that is true. i'm thinking the GPU/CPU and radiator portion of the loop will all be run in series to insure an adequate flow rate through the blocks. (although a parallel CPU/GPU loop has always intrigued me) I'd like to figure out the pump configuration. Perhaps eliminating one pump as Ryan suggested will simplify matters. Toolmaker's data is nice, i just haven't made my through all of it yet.

 

[4Ryan6]

I just noticed something interesting... everyone who's been helping to answer my question also posted in Toolmaker's thread, this must be a pretty tight corner of Tom's Hardware I've found myself in. That's pretty cool, and has given me some perspective while reading those linked threads. Thanks everyone for the help.

If I had the extra pumps, which I actually do, instead of series I would just run 2 independent loops and isolate the DeltaT of the CPU from the GPU or GPUs, many say it doesn't make any difference but from my own testing it does give a slight overclocking edge to the CPU, not to have a constant additional heat load added to the CPU from the GPUs.

That said, my setup absolutely requires running 2 loops as my CPU is chilled water cooled and my GPU is traditional radiator cooled, which is below ambient for the CPU cooling and ambient for the GPU cooling, so my setup requires 2 loops.

Toolmaker_03 has made some good points in his discoveries and my input was hopefully to get him to refine his presentation and present it as an optional cooling solution, not come across as a fact of life as many new to water cooling are barely ready to enter the world of water cooling, much less ready for an additional expense adding a 2nd pump and radiator. (basic buy)

I have tried to relay that cooling like his or mine is specialized cooling and is way outside the everyday water cooler that water cools for one of 2 reasons, either they just love the concept of water cooling period, or they are after higher overclocking head room.

So regarding the thread there's plenty to learn but is the end result worth all the monetary investment that is required to get there?

Anything beyond traditional water cooling is an option left to the user to decide if the ends justify the means.

I've made comments regarding my cooling and if you are curious it is stickied above.

I will still stick to my original suggestion to keep the 2nd as a backup because the pump is the heart of the system and though 2 running may lessen the load on each other, they don't run forever.

 

[toms0033]

Thanks Ryan - I'm going to take your advice because your logic is sound. Also, I was not happy with the dual pump/res mock-up. It was too cluttered. I redesigned where a single pump/res is going to mount, then broke the case down for powder coating. I wound up blasting and coating all the small parts last night. (that corsair paint is absurdly tough to blast through)

Thank you to everyone for your responses, and insight.

 

[4Ryan6]

Thanks Ryan - I'm going to take your advice because your logic is sound. Also, I was not happy with the dual pump/res mock-up. It was too cluttered. I redesigned where a single pump/res is going to mount, then broke the case down for powder coating. I wound up blasting and coating all the small parts last night. (that corsair paint is absurdly tough to blast through)

Thank you to everyone for your responses, and insight.

Post some pics when you get it all finished and back together?

 

[toms0033]

Thanks Ryan - I'm going to take your advice because your logic is sound. Also, I was not happy with the dual pump/res mock-up. It was too cluttered. I redesigned where a single pump/res is going to mount, then broke the case down for powder coating. I wound up blasting and coating all the small parts last night. (that corsair paint is absurdly tough to blast through)

Thank you to everyone for your responses, and insight.

Post some pics when you get it all finished and back together?

Absolutely, honestly I should have started a build log for this one, but I'm already done with the fabrication, and it's been such a drawn out process no one would have wanted to follow it

 

[4Ryan6]

I totally understand some of my stuff has been so drawn out, I just did not want to take the extra time to take pictures during the process, because that time was needed to get it completed and also some of what I've done I did not know 100% for sure it was going to be successful.

 

[toms0033]

I totally understand some of my stuff has been so drawn out, I just did not want to take the extra time to take pictures during the process, because that time was needed to get it completed and also some of what I've done I did not know 100% for sure it was going to be successful.

Agreed! 😀