What benefits do dual pumps provide?

Hey guys,

I have pretty much finished my new build, and I'll be updating the thread that I created back when I ordered the parts here sometime this week with videos and photos.

In the mean time, I'm having to RMA my pump and while I was on Swiftech's website I noticed they just released the MCP35X2. Basically it is two MCP35X's with a dual pump top they've created.

What are the benefits to dual pumps? I know in huge loops there can be too much restriction for one pump, but man that would have to be a huge loop to "out-restrict" a single MCP35X.

Also, I take it redundancy is kind of an insurance against one of the pumps failing?

Why would you normally dual pump? Huge loop? Redundancy? Something Else? All of the above?

Trying to determine if there would ever be any reason for me to buy a second one and use an X2 top.

from what i have read two identical pumps in parallel will increase your flow rate, and two in a series will increase your overall max head (i have only read this and have no experience with dual pumps)
i am planning to buy a second swiftech mcp655 to add in parallel to my loop, when i water cool my ram and nb bcuz the smaller blocks and tighter bends will increase the restriction throughout my loop

as for the redundancy, some people probably do, but imo that would be just an added benefit and not a full on reason to buy 2

Redundancy is why I did it. I had a pump fail and temps went nuts thank god my fan controller (touch 2000) had a tiny alarm I heard.

thebski said:

Why would you normally dual pump? Huge loop? Redundancy? Something Else? All of the above?

Click to expand...

All of the above, including the something else. I don't have a custom loop and have never built one, but have quite a bit of xp in forced convective cooling in power plants. The processes are the same.

In addition to overcoming headloss in a large system and redundancy, sometimes serial pumps are used in order to allow one pump to provide net positive suction head to another larger pump, although I doubt that ever comes into play on something like a custom loop cooler. I guess if the system were large enough it could.

The one thing dual pumps and higher flowrate will not provide is greater cooling. That will be a function of the heat transfer area of your radiator and mass flowrate of air through it. I don't think you were under that impression but I have heard that idea thrown around quite a bit, so wanted to nip it in the bud real quick .

I'm thinking about adding a second for redundancy once I get two water cooled graphics cards in the loop.

It probably doesn't make much sense now to purchase a $130 ($50 dual top, $80 second pump) "insurance policy" to only cool a $300 CPU, but if I've got a $300 CPU and $1400 of graphics cards depending on it, that's a different story.

Like I said I'm having to RMA my pump right now because it has trouble starting (you can find that thread on the topics page as well) while it's brand new. It didn't start one time and it threw the CPU Overheating warning at me and I HATE that. Don't even want to know how warm it got. That's almost got me scared into thinking a second is necessary when you're cooling a lot of $$$.

thobel, I see you're cooling about $3000 worth of hardware with yours so that's a no brainer.

LZ_Xray, I figured the only time it would actually provide a cooling benefit would be if you didn't have an adequate flow rate with only one pump, but I appreciate the confirmation. I'm certainly not in that category, but the rest is something to think about.

LZ_Xray said:

The one thing dual pumps and higher flowrate will not provide is greater cooling. That will be a function of the heat transfer area of your radiator and mass flowrate of air through it. I don't think you were under that impression but I have heard that idea thrown around quite a bit, so wanted to nip it in the bud real quick .

Click to expand...

This is not true, I just grabbed a random skinneelabs article, check it out http://skinneelabs.com/supreme-hf-ni-review/7/. More flow = lower temps...the physics behind this id that there's greater shearing in the boundary layer with faster flow, which induces more heat transfer.

m0r7if3r said:

This is not true, I just grabbed a random skinneelabs article, check it out http://skinneelabs.com/supreme-hf-ni-review/7/. More flow = lower temps...the physics behind this id that there's greater shearing in the boundary layer with faster flow, which induces more heat transfer.

Click to expand...

Note that graph is flowrate vs cpu temp "for 30C water temp". They're maintaining the water temp at a given value to test the heat transfer ability of the TIM. If you artificially maintain a given fluid temp in the system, its true that more flow will always lower temps right down to just above ambient, but that's not what happens in a forced convective system. The graph they're showing is an infinite heatsink. It would be true if you had something akin to a semi truck radiator cooling your cpu, where the heat removal ability of the sink is many orders of magnitude larger than than the heat production ability of the source.

Without the proper symbology, Q=mc(deltaT). Q is the thermal power you're sourcing at the cpu and sinking at the radiator. When you increase the pump flowrate (increase m), the coolant spends less time in the cpu waterblock being heated, and likewise less time in the radiator being cooled. Therefore the deltaT across the cpu gets smaller, deltaT across the radiator gets smaller, the rate of heat transfer (Q) stays the same, and the temp of your cpu stays the same. The only way to dissipate more thermal power and lower the cpu temp is to increase the deltaT at the sink, which is done by either making the radiator surface area larger increasing the speed of the cooling fan. This is why an H100 cools better than an H50 at high cpu loading with the same liquid flowrate.

Laminar boundary layer effects are a factor in high Q, high pressure systems, where nucleate boiling at the heat transfer surface is a primary means of heat removal. A system this size operating at such a low power and high relative flow through the waterblock and radiator has negligible boundary layer effect.

So much wrong with your post I don't even want to cover it, sufficient to say, testing has shown that higher flow = lower temps. Skinnee does not have the gear to adjust water temperature for flow, he was adjusting temps for temps above ambient, since he does not have access to a thermally controlled environment. It's probably not labeled the most clearly, but if you read the methodology, it's made very clear the exact testing process.

You can use formulas and physics all day, but I'm sure anyone who works in engineering like you seem to knows that application trumps theory all day long...you can't argue with application. Also, for the radiators we use to dissipate the 200W or so of heat generated by a CPU, they don't even BEGIN to approach saturation. I think you're STRONGLY underestimating the power of the radiators used for PC watercooling. I don't have the meche/fluid dynamics background to debate it with you in theory (if you wanna talk about EE I can go all day though ), so this is the best I can do...but I've got it on pretty good authority (wanna say that owenator has a masters in either Mech or Nuke E) that the information I presented in my first post is correct.

Provide evidence and this can be a debate. Right now it's not.

"Mech or Nuke E" - Hello navy nuke! I started out there 25 years ago.

LZ_Xray said:

Provide evidence and this can be a debate. Right now it's not.

"Mech or Nuke E" - Hello navy nuke! I started out there 25 years ago.

Click to expand...

Like I said, I don't have the background to debate it with you, I can just point you to the evidence of what testing has shown to be true and allow you to draw your own conclusions.

Um, if flowrate has no effect on heat transfer and subsequent temps, exactly how does larger tubing size give you lower temps when all other factors remain the same??...and before you go to surface cooling of the tubing, we're using PVC tubing in a warm case...
Better yet, prove it to yourself. Kink/pinch a tube and watch your temps rise (on anything. car, computer, boiler, anything).
Water flow just became more important, no?

Water speed in your system is EXACTLY as important as fan speed, and both serve the same function to remove heat from your system. Faster fans=cooler temps, faster water=cooler temps. The only difference is one is on the inside surface and one is on the outside surface.
Since water has higher viscosity and a higher tendency to cling to surfaces, water speed cannot be ignored....awesome fans cannot make up for a poor pump because it's a matter of the heat just not reaching the airflow.
There is a point of diminishing return, because increasing pump power will also mean you are increasing heat dump (the pump is water cooled too), so mega pumps don't often have an advantage in our little systems.

So what say ye now?

***One more point to make, if water flow made no difference to cooling, why didn't Navy engineers use all 1/4" piping in subs to give y'all more room? Hmmm?

BTW, I'm not endorsing dual pumps. I think it's more than is needed, and if you've designed a loop that absolutely needs a dual setup with available stock components, then you've done something wrong.
I'll respect what others' are doing, but please forgive that I won't follow.

Diggrr said:

BTW, I'm not endorsing dual pumps. I think it's more than is needed, and if you've designed a loop that absolutely needs a dual setup with available stock components, then you've done something wrong.
I'll respect what others' are doing, but please forgive that I won't follow.

Click to expand...

Dual pumps is great for redundency if/when one fails

You still have a working system

You Don't need to worry about the system burning down the house while you sleep

You function normal while waiting for the replacement part

I agree most people would never "need" 2 pumps. I view mine as hot swap enabling my system

I can see where you may choose redundancy, but burning down the house isn't going to happen. It takes more than 500°F to set almost anything alight, including flash paper and black powder. By then you've lost the power on and power good signals to your PSU and the system goes off.
Smokeless nitrocellulose gun powder can ignite around 170°C but I'm thinking it would be just silly to make a computer case out of that, no?

Diggrr said:

Smokeless nitrocellulose gun powder can ignite around 170°C but I'm thinking it would be just silly to make a computer case out of that, no?

Click to expand...

Guys, I just had a really cool idea for a mod

That would be the fast way for you to get to Hollywood!

Diggrr said:

I can see where you may choose redundancy, but burning down the house isn't going to happen. It takes more than 500°F to set almost anything alight, including flash paper and black powder. By then you've lost the power on and power good signals to your PSU and the system goes off.
Smokeless nitrocellulose gun powder can ignite around 170°C but I'm thinking it would be just silly to make a computer case out of that, no?

Click to expand...

I'm assuming you mised the smiley face..

No I saw it. Just in a mood today. Snowed again last night and I am cranky!
3 more years until I can get out of Yankee territory for good!

LZ_Xray said:

The one thing dual pumps and higher flowrate will not provide is greater cooling. That will be a function of the heat transfer area of your radiator and mass flowrate of air through it. I don't think you were under that impression but I have heard that idea thrown around quite a bit, so wanted to nip it in the bud real quick .

Click to expand...

Turn the flow down to nearly nothing what happens? Cpu temps will go up by 10+C. Turn the flow (pump) off, and what happens, cpu temps will be 98-100C and the cpu will be throttling.

The water is a transfer medium, it carries heat from cpu to rad. The faster heat gets from cpu to rad, quicker it can be dissipated to the air.

So at zero flow or very low flow cpu temps rise dramatically. At what flow rate does thermodynamics become irrelevant, ie at what flow rate are you stating that cpu temps no longer benefit from any higher flow?

I found this thread searching for dual pumps one loop as I was wondering about this. i have a mcp655 that is about 3 years old (guesstimate) and it's been running 24/7 for a large portion of that. So I am a little worried about it dying and burning something up.

I'm not real worried about my cpu but I have a 6970 that was fairly expensive. I know the cpu would throttle when it overheats (would it shut down in complete coolant loss?) but I don't' really know about the vid card.

I'm also about to be pushing water through a gtx360, swifttech 240, cpu block and EK fullcover gpu block and am slightly curious about the strength of the pump to flow adequately.

Should I really be concerned? Is two pumps a really smart idea? Would serial or parallel be better if I did go that route?