Dual water pump?

[Richard]

I know it might seem a bit redundant, but what the heck.

Would running two low GPH water pumps yield any benefit over one higher rated water pump? For one thing I can see the benefit of having a backup in case of a failure, but I was also thinking that maybe a dual pump system would yield "truer" flow rates?

Perhaps, placing one pump before the radiator and one after the radiator? (Since most of the convolutions are in the radiator) Would the second pump take up the slack of the first pump? Or would the flowrate remain pretty much constant?

A lot of questions, I know. However, I'm thinking of building a "failsafe" water cooled rig. I'm more or less trying to see if there are any real negatives associated with such a contraption.

Thank You for your time.

 

[DamnFast]

Putting two pumps won't increase the flow much. It may even reduce the flow possibly. And as far as the failsafe is concerned, if one pump fails, the water flow will be restricted so much since the fins in the pump would be another big friction that the other working pump has to go through. So, I don't think having two pumps is ideal for fail safe. The best failsafe feature is the temperature monitor of any sort. i.e. MBM or any probe. Just set the warning temp at high 40's and shutdown temp at high 50's. This way, it will protect the cpu from burning in the case of pump failure and will also save the water pipes from melting.

 

[Richard]

Thanks for giving me a reason to think it through a little more carefully.

However, I think I may have come up with something else.

Still keeping with the dual pump theme, but potentially heading off the problem of one pump shutting down and causing a blockage in the system. I came up with a method that would incorporate two pumps basically independent of each other, but still working together.

Rather than place the pumps far apart. Keep them on the same side of the radiator.

Here's my idea.

Place a "Y" in the line going to CPU. Since water will travel the path of least resistance it will flow toward the CPU. Then place another "Y" immediately after the radiator.

This would effectively isolate one pump should the other fail.

What do you think?

DamnFast (Feb 27, 2001 08:50 p.m.):
Putting two pumps won't increase the flow much. It may even reduce the flow possibly. And as far as the failsafe is concerned, if one pump fails, the water flow will be restricted so much since the fins in the pump would be another big friction that the other working pump has to go through. So, I don't think having two pumps is ideal for fail safe. The best failsafe feature is the temperature monitor of any sort. i.e. MBM or any probe. Just set the warning temp at high 40's and shutdown temp at high 50's. This way, it will protect the cpu from burning in the case of pump failure and will also save the water pipes from melting.

 

[DamnFast]

Richard999 (Feb 27, 2001 09:36 p.m.):
Thanks for giving me a reason to think it through a little more carefully.

However, I think I may have come up with something else.

Still keeping with the dual pump theme, but potentially heading off the problem of one pump shutting down and causing a blockage in the system. I came up with a method that would incorporate two pumps basically independent of each other, but still working together.

Rather than place the pumps far apart. Keep them on the same side of the radiator.

Here's my idea.

Place a "Y" in the line going to CPU. Since water will travel the path of least resistance it will flow toward the CPU. Then place another "Y" immediately after the radiator.

This would effectively isolate one pump should the other fail.

What do you think?

It may be possible. However here's my thought.
Well, that's going to lead to another problem. Having two Y connectors in the water line will lead to additional friction and turbulence might slow down the overall flow. And also water will not go through the Y connector in 50-50% exactly. They will take whichever is shorter and flatter and the majority will take the easier route leaving one pump not working at it's full potential.
You can give it a try and tell me the results though.
Good luck.

 

[dimmreaper]

I agree that putting two pumps in a series will not help the flow-rate much, but I think it may give the pumps more head (the ability to lift water further in a verticle manner).

I also agree that putting two pumps in parallel is not a very promising solution.

www.leufkentechnologies.com has some nifty new in-line thermistors, placing one of those close to the waterblock could serve as a back-up temp alarm. Set MBM to shut down if the CPU temp rises above 50C and set it to shut down if the watertemp rises above say 35C. Now that is a fail-safe system!

 

[Richard]

I think I have an idea of how I will build it now.

To cut down on the number of "Y" connectors I will build a single reservoir with 2 submersible pumps. I'll place the pumps in side by side with the return nozzle directly centered over the return inlets.

You make a good point about the "Y" adding turbulence. I work in open hearts and I use a "Y" connector that is designed to reduce turbulence over standard "Y" connectors. Promoting laminar flow is very important in on bypass procedures; so I may have to borrow one of these in my experiment.

In an attempt at increasing efficiency I'll try and keep the tubing as even as possible. I don't think I want to place the "Y" directly at the exit as the high pressure head will surely cause issues. Maybe, a 10 cm length then the "Y" would facilitate laminar flow?

I have a feeling I'm going to be trying several different combinations.

The only thing I'm not sure of is what rating pumps to use. Should I use something > or < 100 GPH? I'm thinking of going the lesser route.

Criticisms and comments welcome!

 

[Richard]

Very good idea. An inline thermistor. I like it.

Ultimately, my goal is multi-faceted. I want to build a system that will give me peace of mind.

I've watercooled in the past, but always gone back to air cooling simply because I didn't trust the apparatus.

I'm a strong believer in having a contingency plan.

We'll see how it all works out. I'll take plenty of pictures when the project is complete.

Jeff Evans (Feb 27, 2001 10:25 p.m.):
I agree that putting two pumps in a series will not help the flow-rate much, but I think it may give the pumps more head (the ability to lift water further in a verticle manner).

I also agree that putting two pumps in parallel is not a very promising solution.

www.leufkentechnologies.com has some nifty new in-line thermistors, placing one of those close to the waterblock could serve as a back-up temp alarm. Set MBM to shut down if the CPU temp rises above 50C and set it to shut down if the watertemp rises above say 35C. Now that is a fail-safe system!

 

[dimmreaper]

I just had an Idea that might be what you need. Use dual submerged pumps like you expressed interest in. But instead of using a "Y" at all, build a waterblock with two inlets. Use 3/8" ID tubing for the waterblocks inlets. Use 1/2" ID tubing for the return line (or "outlet"). This should provide some serious flow, and give you piece of mind friend.

 

[Richard]

I spoke with our perfusionist. (She runs the heart / lung machine.) Man, you guys would get a kick out of these pumps. The pump is a combination gravity and centrifugal pump.

She offered quite a bit of insight into the fluid dynamics of the setup, and affirmed the importance of eliminating as many connections as possible.

I now have the plans laid out for the system and have ordered the parts. I have a good feeling about the design, and hope other water cooler enthusiasts can get some ideas from my efforts.

Hopefully, the rig will be complete sometime next week.

 

[NUTNDUN]

Hey Jeff,
I tried getting ShutDownNow to work with MBM but I couldn't figure it out. Anyone feel free to let me know.

 

[Richard]

Jeff, thank you very much. I currently have a Copper Overclockershideout waterbock.

However, I liked your idea so much I've contacted DangerDen and asked them to put a 1/2" connector acting as an outlet, to their Maze waterblock.

Should be an interesting experiment. If anything I'll have a fairly unique setup.

Jeff Evans (Feb 27, 2001 11:14 p.m.):
I just had an Idea that might be what you need. Use dual submerged pumps like you expressed interest in. But instead of using a "Y" at all, build a waterblock with two inlets. Use 3/8" ID tubing for the waterblocks inlets. Use 1/2" ID tubing for the return line (or "outlet"). This should provide some serious flow, and give you piece of mind friend.

 

[William]

actually, with the bigger the pipe, the slower the flow rate. In a closed circuit system, the only way you are going to incerase the flow rate is to put in a bigger pump or rev up the pressure on the outlet, and rasing the pressure is bad. In a bigger pipe vs a smaller pipe, they transfer the same amount of water, just the larger one is going at a slower rate compared to smaller one which moves at a much faster rate. The formulas is the following.
A=Area
V=Velocity

A1V1=A2V2

all numbers should be subscripts. Of sorse this is assuming an ideal fluid with no turbulence or viscosity, but it does gives you the appropriate demonstration of what happens with larger pipes.

 

[MEfreak]

Concerning 2 pumps inline: inline pumps raise the total head of the system. They do not raise the flowrate significantly in a closed system. If you have an open system and the pump can't provide enough pressure, a second inline pump will double the pressure, which *will* improve the flowrate. You need 2 pumps in parallel to increase the flowrate. You wont double it; maybe you will increase flowrate by 160-170% depending on the pump.

One situation where inline pumps help is if your hose barbs are too small for your system and they pinch the flow. Say if your barbs are 1/4" and your tubing is 1/2", the barbs will add huge pressure losses to the system. But running 2 pumps inline can overcome the added head. I like this system when using water blocks with multiple small diameter channels. More surface area is possible, better cooling is hte result along with increased pressure requirements. I have a Y split before the block and 2 small channels in the block itself.

Concerning connections: Dont worry about y connectors too much. If you go for the low loss designs, they won't hurt system performance too much. You won't be able to tell the difference if your pump is up to spec. Some of the el cheapo brands will suffer: solution is to buy a descent pump. Look for pumps that will maintain high flowrates at high head pressure. If you use the parallel pump setup mentioned below, you can use two cheap *high* flowrate pumps to blast your system and not break the bank. As far as T connections are concerned, aviod them like the plague. And hose barbs, watch out for those guys. They will sap your pressure if you use too small connectors (like 1/4 in a 3/8 system).

Here is another idea for all of you. Slow the water down in the radiator. Use 2 parallel radiators or one radiator with larger tubes. This won't help if your current radiator is already cooling the water to room temp. But if your radiator is flowing the water too fast, it won't be able to cool the water as effectively.

Another idea. Use 2 pumps in parallel with check valves. This way you can have a backup. The check valves will prevent water from turning back to the failed pump at the y connector. When both are running, you get increased flow rate. When one fails, you should still get enough flowrate if your backup pump is up to spec. Hell, throw a switch on the backup pump to switch it on when the first pump goes bye bye (put a LED on your case to indicate pump failure).

Anyone else with some ideas?

 

[dimmreaper]

dunno260 (Mar 02, 2001 02:22 p.m.):
actually, with the bigger the pipe, the slower the flow rate. In a closed circuit system, the only way you are going to incerase the flow rate is to put in a bigger pump or rev up the pressure on the outlet, and rasing the pressure is bad. In a bigger pipe vs a smaller pipe, they transfer the same amount of water, just the larger one is going at a slower rate compared to smaller one which moves at a much faster rate. The formulas is the following.
A=Area
V=Velocity

A1V1=A2V2

all numbers should be subscripts. Of sorse this is assuming an ideal fluid with no turbulence or viscosity, but it does gives you the appropriate demonstration of what happens with larger pipes.

When we refer to "flow-rate" we are not talking about fluid velocity. We are talking about a volume unit of fluid going from point A to point B in time unit. We are talking about gallons per and hour in these cases. You can pump a lot more gallons an hour through a 7/16" hose than you can a 3/8" hose. That's just common sense.

 

[William]

of course