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

Redundant Array of Inexpensive Pumps

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

RonnieG

Member
Joined
Dec 3, 2002
Location
Philippines
RAIP (redundant array of inexpensive pumps) :D Cheesy name for my new project, basically 2 really cheap pumps in serial. Not really an array, but definitely redundant.


raip1.jpg



Those are Resun SP-600, the same brand Volenti is using. I think he's using the SP-1200 model. The difference between Volenti's and this is I'm using only 1 "layer" of acrylic, to keep the parts count and complexity down (at the expense of a little performance though...). Kinda like a russian firearm...

At first glance it may seem parallel but I assure you it's serial :)

The rationale behind this project is redundancy, NOT performance, hence the weak pumps, and only 2 at that. I also want the impellers visible from the outside, and perhaps some blue LEDs in the right places.

I'll be adding some magnetic shielding also.

I'll update on developments...
 
RAIP! RAIP! RONNIES RAIPING!


naw interesting idea

that seems like... it would be parrallel, both drawing from the same channel, both... putting into the same channel... seems parrallel to me...
 
instead of getting to the bells and whistles you have to first be true to your name. You cant run both pumps at the same time all the time, that's not very redundent, that's just running them in parallel and hoping if one dies it isn't something that will kill both. Are you going to rig up a switch that activates the relay to the second pump when water pressure in the output of the first pump falls below a certain force ?

A redundent pump setup would only be really valuable if you had a single pump that does all the work and then if it fails, activates a backup pump that works like new, designed to do it's job and alert the user that a problem has occured and the system needs immediate attention.

What you have pictured above looks like a simple parallel pump setup despite your describing it as series. It looks like the input is a Y split going to both thus it cannot be in series. Series would have the second pump's input be the first pump's output and then their ends are attached to the rest of the system, no splitters. A parallel setup is always redundent in the way that the two pumps can run independently. Having a redundent setup in series would be foolish. If one pump clogs the second pump is screwed.

Anyways, it'll be a snazzy parallel pump setup if you continue on with your above plans. But it wont be any more redundent than any other parallel pump setup i've seen.
 
I knew you're going to get confused...

Here's a new pic:

raip2.jpg


I'm going to seal the area marked RED. Does it make more sense now?
 
A lack of information is not getting confused, it's just not being telepathic. You never even mentioned doing anything like blocking one off. And how does it unblock itself? you never mentioned your mechanisms for enabling the second pump on first pump failure, you did, however, talk about putting blue LED's on it and using plexi so you could see the impellors. What are you going to do to make it redundent and not just a parallel pump setup?

Are you going to use a digital pressure guage to determine when it's necessary to activate your redudent pump? your valves dont need to be wired up to anything, simple one directional valves can work fine, and the redundent pump's own pressure will open them when activated, it's just the matter of turning the redundent pump on that is of main importance. forget the windows and leds and shielding. How are you going to do that - that's the interesting thing.
 
safemode said:
instead of getting to the bells and whistles you have to first be true to your name. You cant run both pumps at the same time all the time, that's not very redundent, that's just running them in parallel and hoping if one dies it isn't something that will kill both. Are you going to rig up a switch that activates the relay to the second pump when water pressure in the output of the first pump falls below a certain force ?

A redundent pump setup would only be really valuable if you had a single pump that does all the work and then if it fails, activates a backup pump that works like new, designed to do it's job and alert the user that a problem has occured and the system needs immediate attention.


Nah I want to run BOTH pumps at the same time, the extra performance wouldn't hurt :D

And no, serial pumps won't choke if one stops. Try it for yourself.

Oh you forgot to mention the health monitoring for the pumps. Temp sensor, odometers, tachometers. Plus their own smart microprocessor-driven electronics with a database of all past characteristic failures stored in the flash memory. This will allow predictive evaluation of impending pump failure.

Maybe you could also try load balancing the pumps. If one pump falls below threshold, the other varies its duty cylce to maintain the total flow rate. This is to preserve the "life" of the second pump, as well as utilise what's remaining of the first.

Redundant enough?

What about remote alerts via SMS? There are some off-the-shelf building-blocks for this. Receive notifications of pump failure right on your mobile phone. Pretty cool huh...

Redundant Internet pumps. Watch those little performance graphs from anywhere in the world with it's web interface. Even lets you assign which pump is active....

Or self-repairing nanotech pumps... :D :D :D
 
Last edited:
safemode said:
A lack of information is not getting confused, it's just not being telepathic. You never even mentioned doing anything like blocking one off. And how does it unblock itself? you never mentioned your mechanisms for enabling the second pump on first pump failure, you did, however, talk about putting blue LED's on it and using plexi so you could see the impellors. What are you going to do to make it redundent and not just a parallel pump setup?

Are you going to use a digital pressure guage to determine when it's necessary to activate your redudent pump? your valves dont need to be wired up to anything, simple one directional valves can work fine, and the redundent pump's own pressure will open them when activated, it's just the matter of turning the redundent pump on that is of main importance. forget the windows and leds and shielding. How are you going to do that - that's the interesting thing.


Sorry this isn't the mod that would interest you. Try looking elsewhere, or Google.

KISS, keep it simple stupid
 
RonnieG said:
Nah I want to run BOTH pumps at the same time, the extra performance wouldn't hurt :D

And no, serial pumps won't choke if one stops. Try it for yourself.
It was if one gets blocked and stops. Not if one just stopped because it felt like stopping. If the path of water is blocked then water cannot pass on. What is confusing you about that? Serial is not what you are doing Period. Look up serial setup somewhere, you have it confused with something else.

Oh you forgot to mention the health monitoring for the pumps. Temp sensor, odometers, tachometers. Plus their own smart microprocessor-driven electronics with a database of all past characteristic failures stored in the flash memory. This will allow predictive evaluation of impending pump failure.

Maybe you could also try load balancing the pumps. If one pump falls below threshold, the other varies its duty cylce to maintain the total flow rate. This is to preserve the "life" of the second pump, as well as utilise what's remaining of the first.

Redundant enough?
Sad. actually.

What about remote alerts via SMS? There are some off-the-shelf building-blocks for this. Receive notifications of pump failure right on your mobile phone. Pretty cool huh...

Redundant Internet pumps. Watch those little performance graphs from anywhere in the world with it's web interface. Even lets you assign which pump is active....

Or self-repairing nanotech pumps... :D :D :D

Or you could stop pretending you're building a redundent pump setup and call it what it is. A parallel pump setup with some lights and windows.

A redundent pump setup would not be difficult to construct if you wanted to. You would just need a pressure guage on the output line of the primary pump wired to trigger the power relay for the secondary pump. You'd use one directional valves on the input and output of the secondary pump to make sure there is no flow through that pump while it is off. Wiring wise, you could get away with not really even altering the pressure guage, a little metal contact paint on the needle and two well placed pins perpendicular to the needle attached to a switch can activate the switch once it's contacted once, all subsequent contacts and such wont matter as the switch ignores all further contact. switch activates relay and relay works pump. the only complicated part is the switch. and there are numerous ways to handle that.
 
RonnieG said:



Sorry this isn't the mod that would interest you. Try looking elsewhere, or Google.

KISS, keep it simple stupid

nothing wrong with kiss. But what you're doing is a simple parallel pump setup. It's not what you originally suggested it was by the thread name. So yes it's disappointing to find out what it really is.
 
Sorry safemode but this is a serial setup.

Just like you, I initially thought it was a parallel setup because, you were right about that, we missed a crucial information.

Then after the second picture, with the sealed area, I understood how it's indeed serial. You asked how the seal unblocked itself. I think you assume the first pump works alone and in case of failure, the seal unseals, allowing water to go to the second pump in what would be a parallel setup. but this is not how it works. The seal is actually permanent. No water goes through it, ever. Look closely at the sealed channel to the upper pump: right after the seal it's drilled down to the first pump's outlet. Here is how I understand it:

raip3.jpg


Definitely serial, eh?

And before you bark at RonnieG for insufficient info, keep in mind that I understood it worked that way with the info he provided here. Maybe a little reading between the lines was needed but all the info was there. When I first looked at the first picture, I was like "this is parallel, not serial". Then I read RonnieG's comment that even if it looks parallel, it's not. He was aware that it looks parallel. So instead of going "No freakin way!! This is parallel!!" I just assumed RonnieG was not the idiot I first thought he was ;) and that there was something I didn't know about the design. I was curious about it and then, with the second picture, I learned all the missing info I needed.

Hope that settles this little conflict ;) Let's resume with a discussion of the specifics.

Edit: Oh, and I think that at the light of this the RAIP name is appropriate. Serial pumps are definitely redundant, aren't they?
 
Last edited:
RonnieG said:
And no, serial pumps won't choke if one stops. Try it for yourself.
This is one of the specifics I'm interested in. I never tried for myself (still didn't do the leap to watercooling, actually). From the design of a centrifugal pump, it seems like, indeed, water can flow through it even if the pump isn't running. But it also looks like it will still reduce the flowrate.

One question about that: what happens to a pump if it doesn't work, let's assume it's unplugged for example. Is the spinning wheel stuck in its position, or is it free? Will the water have to go around the wheel parts like it's a fixed object, or will the waterflow make the wheel spin, thus reducing the resistance?

Actually, as the water enters at the middle of the wheel, I'm not even sure it would make it spin if it was free.
 
Hey I like this project and can't wait to see it finished. I hope you show some pix of it's output with no tubing on it. One with both pumps going and another with just one. I'm wondering if single output will be the same regardless of which single pump is doing it.

And safemode, I think he used the term redundant mostly for a play on words with the RAID/RAIP thing, I kinda liked it myself 8P~

peace.
unloaded
 
as i said above (and why i insisted he really meant parallel)

The serial setup assumes one thing will not happen that is very common among pump failures and that is that the pump clogs. If the pump clogs how is water going to pass through it ? You assume you still have a clear pathway for water when the primary pump fails and what is to stop the secondary pump from failing as well due to buildup since they are both always running ? A serial setup makes very little sense as a redundent safety system for pumps. Is the only type of pump failure you are thinknig is going to happen is that the impellor cleanly stops spinning? If anything prohibits the movement of water in either of the pumps then the whole system is stuck. Where is the water going to come from if it's blocked in one of the pumps?

I was assuming ronnie was not an idiot and was simply mistaken as to the meaning of series and parallel because I see a serial setup as working against the intended purpose of the namesake due to it's great weakness in function. I dont see how you can say sometihng is redundent while depending on the very thing it's supposed to be rudundent against. A redundent system is made of equally functioning parts that are as independent of one another as possible to ensure the greatest chance that one works if one fails since the likelihood of the same thing affecting both the same way are greately lessened. A serial setup is a mistake if your purpose is redundency and why you wouldn't see a parallel setup as the basis for making a system redundent is kind of strange.

Forget impellors blocking the way of water by not moving for whatever imagined reason. Throw in a calcium buildup that is common in many pump failures and try moving water through it. by definition of a serial setup the only way the water is getting to the second pump is by flowing through the first and if the water cant flow through the first then the second is left high and dry doing nothing. By definition this sucks as being a redundent setup. A redundent setup is supposed to work still if one pump fails but in the case of a series setup this isn't the case. A parallel setup makes much more sense and works much to the advantage of setting up redundent pumps. A parallel setup is already redundent by just having it, but it's not the most efficient use of redundent pumps because you're using them both all the time. But that gets into the above emails where i thought he was saying one thing that was wrong and building another that was on the right track. Now you clarify things and it's was saying one thing that is right and building something that is on the wrong track.

the thing that got me the most was that he even quoted KISS yet ignores the most obvious and easy to construct redundent system in favor of more work and a system that has a fatal flaw by it's very nature. KISS is not short for going out of your way to make what could be at least a functional system into a flawed system requiring work to do so. I mean he has what could be a parallel setup but is going to alter it and block a tube and drill in other places so that it becomes serial. That makes no sense from a kiss point of view or a redundent design point of view.

hopefully the plexi hasn't been altered yet, maybe there's time left to salvage some of the project's purpose.
 
Sure it's gonna be a bit restrictive, changing directions like that in the pump's outlets. But when it gets down to it, most commercial waterblocks are lucky to allow 60-100 gph through using a conventional magdrive pump. If he can push at least that, then one pump failing and one surviving IS redundency, even in a serial setup. Even if the waterflow is slowed to a trickle, the cpu's gonna live.
There's no real need to over-engineer with electronic fail-safes and such. That's whooping the problem with a much bigger stick than is needed.

If your pumps are clogging, quit bathing your cats in your bong.
Calcium buildup? I have never seen that using distilled water and the appropriate additives...and yes, I've been doing this for a while, more than many here.

Freedom of speech is NOT an invitation to jump into any thread you like and start pi**ing all over the place, and four stars worth as alot of pi**ing....Post positive, I dare you!
 
lol diggrr,

I like this kind of mod! its pretty interesting! although there is a lot of flow direction changes, but it shouldnt be bad at all, Its very interesting project you have going here!
 
I remember vollenti's project similar to this, he had to use some kinda flaps/valves to prevent backflow. I know yours is desgned quite a bit differently, I'm wondering if you'll need similar?
 
nope, he wont, It looks like his pump spins one way because of the way the outlet is,
the reason he had dual outlets was so if it spun the other way it would go out the other side and not suck in the one,
 
I've been thinking about anti-backflow valves too when safemode mentioned parallel redundancy as more sensible than serial. If one of your pumps dies in parallel because, say, it's unplugged, then nothing prevents water from flowing through it the wrong way. The other pump(s) might be pumping through a loop that never goes through the waterblocks.
 
redPEPPER said:
I've been thinking about anti-backflow valves too when safemode mentioned parallel redundancy as more sensible than serial. If one of your pumps dies in parallel because, say, it's unplugged, then nothing prevents water from flowing through it the wrong way. The other pump(s) might be pumping through a loop that never goes through the waterblocks.

If you use a simple Y splitter to run the pumps in parallel there is almost no way that any flow is going to go backwards through the dead pump to create a circular circuit.

The force of the incoming water should be enough to prevent anything like that from happening if your piping is designed with the proper flow direction in mind.
 
Back