• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Pumps in serial connection or dual loop?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Taller51169

Registered
Joined
Oct 4, 2011
So im going to make an upgrade soon by adding another pump/rad in my system and I would like some opinions here if possible. Is it better to have the pumps in serial and thus having 1 big loop with twice the head pressure or make 2 seperate loops? To my knowledge having bigger head will give better results in almost every situation except on full load but i would like to have some second opinions. Here is the loops I'm planning

Loop 1: Pump - rad 120.3 - rad 120.2 - ek fc 480 gtx - ek fc 480 gtx
Loop 2 : Pump - rad 120.2 - rad 120.1 - ek supreme

Or 1 loop with pump in serial
Thanks for help any info is appreciated guys.
 
How many of these radiators do you already have? I have to say that seems a bit excessive even by some of the most stringent standards.
IMO it makes life easier only having to deal with one loop and all my WC has always gone that route. Putting pumps in series will increase head pressure and help overcome the resistance of the loop which is in general better than a high flow configuration since without the head pressure you wont get those flow numbers anyway.

What CPU are you cooling that needs 120.3 of private loop? I have to imagine even with 2x GTX 480s that your tripple and double radiators should handle it without the 2nd double and single. Of course your welcome to add as much radiator as you feel the need for but you might consider scaling it down a bit.
Also what pump are you already using? if you run multiple pumps in the same loop you really want to be sure they are the same pump.
 
Hi, thanks for the fast reply. I was running a single 120.5 loop till now cooling my phenom 810 and a single 480 gtx. I was getting 59 c on the gtx at full load and 30 c on the cpu. That is after 3 hours of furmark/prime 95 all running together. Now that i added my second card on the loop i figured out through some research that the 2x 480 would need a good 120.5 setup in order to perform well. Im targeting to get a dt of around 7c with an ambient temperature of 32 - 34 c. I am also upgrading to a bulldozer now so a good 120.3 for it should be more than enough. Keep in mind that these rads are not in optimal position in the case also. All rads are cooled with scythe kamas at 1800 rpm. Besides all that i was thinking the parallel conection also for the extra safety in case 1 of the pumps die in duty
 
Could just be me but it seems like that temp is a little high on that GPU, but could just be that I havnt used a card quite as hot as those 480s.

If your set on adding more radiator Id drop that 120 single and just add the 240 into the single loop. You might want to add the GPU first and see how it performs before going through the trouble of adding a radiator.

If your concerned about redundancy in your pumps then parallel would be the way to go, it wont add much on overall loop flow but it gives you a fully capable unrestricted backup incase you need it. You pretty much need a res for this type of setup though.
 
Ohh well, 480 gtx on stock cooling idles at 47 c and on max load it reaches 100 c after 2 - 3 hours of play so working it at 59 it aint that bad. Anyway i think ill drop the safety for performance and do a serial.
 
I agree with ssjwizard - probably too much radiator. You should be fine with the 120.5 you currently have unless you *really* want to add more. Pumps should definitely be in series.
 
Parallel won't lead to any added safety or redundancy. You'd have one pump per loop. If one of the pumps fails, the components on that loop will get no flow. In a serial setup you two pumps per loop, so if one pump fails flow rate drops but doesn't stop entirely. This is much safer.

With as much rad as you're planning to throw at this, serial should work out nicely. I'd only go for the dual loop option if you were planning on tolerating a higher dT for the GPU loop. Since you're targetting a dT of 7 deg C for the whole thing, go serial and save yourself a tubing nightmare with all those rads plumbed in.
 
Parallel won't lead to any added safety or redundancy. You'd have one pump per loop. .

We(rather I) were actually discussing parallel pumps, one single loop but with divided pump load. Several ways that could be plumbed in. This would provide (marginally) better flow, and provide redundancy. Series does the same thing BUT it increases flow more while having less redundancy.

In a pump failure situation with parallel your flowrate wont change much at all since the pressure in the loop is fairly static. In a series configuration you increase pressure with both pumps running but if one pump dies its a big dead load of resistance inside the loop now.

Both situations your protected from a total meltdown in case of a pump failure but its a trade off on whether you want more redundancy or more performance. Its kind of like raid 1+0 vs raid 0+1. Both do essentially the same thing but they make a similar trade off.

Either way a single loop for all intents and purposes is the preference of most builders.
 
In a pump failure situation with parallel your flowrate wont change much at all since the pressure in the loop is fairly static. In a series configuration you increase pressure with both pumps running but if one pump dies its a big dead load of resistance inside the loop now.
If you have a single loop with parallel pumps, so that it goes something like -----<=>----- where the '=' represents the two pumps, what about backflow if one pump fails? The path of least resistance for the second pump is then (potentially) back through the failed pump instead of moving through the rest of the loop.
 
If you properly plumb everything it should take a near reversal of direction to do that, add in the resistance of the dead pump and its fairly immaterial. It will change the flowrate some but not as much as in a series. Beyond that inline backflow prevention fittings exist if you were really worried about it.

Anyways we have gone way off topic now. OP if you have any more questions just let us know.
 
I misunderstood your intent, but my point still stands.

In a pump failure situation with parallel your flowrate wont change much at all since the pressure in the loop is fairly static. In a series configuration you increase pressure with both pumps running but if one pump dies its a big dead load of resistance inside the loop now.

Urban legend, Martin tested it a long time back.

Bottom line….Use Series setups in multiple pump configurations, do not use parallel pump setups. There is really no advantage to using parallel over series, series is always better for watercooling.

http://martinsliquidlab.org/2011/04/26/pump-setup-series-vs-parallel/

If you flip through and look at the charts, you'll see that series results in better redundancy (less flow loss when one pump fails), more flow at head pressures typically seen in a water cooling loop, and better pressure at most flow rates. Series beats parallel in every metric in those tests, and they're the only tests I'm aware of that looked at that setup. I'd love to see more data if it's out there.

Parallel only starts to lead the flow race off the right hand side of the chart, in an area of restriction that isn't really likely in a water loop. Even if it were, the gains would be terribly small since flow would already be very high. In the system we're talking about here, it's certainly not in an "extreme low resistance" type of category, so series is his best option if he's running two pumps.
 
Last edited:
So guys, after some more research and with your kind info ofc :) i decided to go down the serial path. I just wanted to ask tho, the normal way to do a serial setup is by going from pump 1 outlet to pump 2 inlet and then on the loop but what happens if i place the pumps splitted into 1 loop. EX Pump 1 - rad 1 - rad 2 - WB - pump 2 - rad 3 - rad 4 - WB. Does it have the same benefits as the pump 1 - pump 2 setup?
 
No it actually doesnt scale that way. I dont have hard date but I recall somone doing tests on this and adding a 2nd pump part way down the loop made very little difference while plubming one into the other made a significant difference.
 
Hmm, i guess by doing a split setup you just earn a bit more steady flow. Thing is tho, i have 2x MCP655-B with EK Res tops on both so i cant actually connect the pump 1 outlet directly to pump 2 inlet. Any suggestions here?
 
It shouldnt matter where in the loop the pumps in series are they will still be in series and stack (double for same pump) head, and in a restrictive loop will significantly increase flow.

Easy way to test this oneself is with a bucket and a gallon jug and stopwatch (time takes to fill a gallon), and alternating loop order.

Martins pic here explains series (vs parallel) well. Series doubles head. The flow however will not be more than single pump at 0 restriction. Follow graph of his high restriction loop and very low restriction.

In my loop, typical restriction with 1 full cover gpu (dual gpu) and cpu block, I got about 1/3 more flow adding 2nd mcp355 in series (measured on components from 2 years ago, had 3 different tops cant remember which one I used).
 
Ye I'm aware that 2 pumps in series = 2xhead pressure and flow steady while 2 pumps in parallel = 2x flow and head pressure steady. The question is, do I have to place the 2 pumps after each other ( pump 1 outlet to pump 2 inlet) or I can split them in the loop and still get the extra head pressure?
 
2 pumps will result in a great flow rate and very very good head pressures. Mount them in a manner that allows for easiest tubing routing IMHO, regardless of the order. I'd like to see the info ssj refers to, as it doesn't make a ton of sense for a closed loop. I guess putting the pressure line of one pump directly into the suction line of the other would be ideal, but in a closed loop everything is directly connect. In any case, I doubt the difference in head pressure/flow would show itself in any significant way (maybe 1 deg? I have no actual data), and isn't that what really counts?

I can appreciate the difficulty in mounting two MCP655s, they're big. A few thoughts spring to mind if you're dead set on getting them connected directly. One is to turn one of the pumps 90 degrees on its side so that the inlet is on top and the outlet is horizontal, motor facing down. That should allow an easy run to the inlet of the second pump, but you'll need to figure out how to hold that first pump in position.

If you've got the budget, there's this thingie from BP.

This method seems to be the simplest, but I guess you don't have the room for it?
 
Ye I'm aware that 2 pumps in series = 2xhead pressure and flow steady while 2 pumps in parallel = 2x flow and head pressure steady. The question is, do I have to place the 2 pumps after each other ( pump 1 outlet to pump 2 inlet) or I can split them in the loop and still get the extra head pressure?

You can do pump rad pump cpu or pump pump rad cpu, the flow will be the same.

Most will do res - pump -pump for filling/bleeding ease. If you do res pump - rad -pump for example, just leave the second pump off, and only fill/bleed with pump after res until most air gone, so second pump doesnt run dry.
 
Last edited:
2 pumps will result in a great flow rate and very very good head pressures. Mount them in a manner that allows for easiest tubing routing IMHO, regardless of the order. I'd like to see the info ssj refers to, as it doesn't make a ton of sense for a closed loop.

Im looking for it, I came across a write up (@XS I think) about 2 months ago when I was researching a bunch of the dual tops and what I found was that 2 pumps directly piped into eachother is the most beneficial, a dual top comes in 2nd and splitting them up made very little difference.

When/if I find it again ill come and edit this post with a link.
 
^ I'd be interested in seeing that if you find it. My intuition would be that it doesn't make a difference where the second pump is, but pumps can behave very differently depending on what exactly is upstream of the inlet.
 
Back