Getting WW, Need help with Flow Design

Okay, so I am getting a White Water block. I really like the single inlet dual outlet idea. I have a question though. In my system I also have a GPU water block. The problem is that it is very restrictive because of the 90 elbows and the size of the barbs. These can and probably will be changed but there is no amount of changing that will allow it to flow as good as the WW and I don't want my CPU to suffer from bad flow caused by the GPU.

So after seeing the picture my question is this. Would the second manifold cause a backup of pressure therefore killing some of the flow "From" the WW? Can someone tell me something better?

Please try to comprehend my extensive, highly detailed, 3D diagram below.

The linky doesnt work. Also, The WW isn't that flow friendly due to its slit design. It does perform better than other WBs out there but not nevessarily the most flow friendly block.

Sorta dont understand your question? can you rephrase it with the diagram working

Ok the picture works... I see that you are planning to go with 2 pumps. Run them in serial with less pressure feeding the high pressure. Why did you decide to use manifolds?

edit
Ok if anything do this... Run the pummps in series which gets rid of the first manifold. And cutting the flow on the GPU by splitting it doesnt look like it would help. In fact I think you might be hurting the flow more that way. So go directly from CPU to GPU becaseu the water goes there anywayz which is restrictive so I doubt splitting will be beneficial.

ILikeMy240sx said:

Ok if anything do this... Run the pummps in series which gets rid of the first manifold. .

Click to expand...

I have heard before that running two pumps in series is not the best setup when both pumps flow the same amount?

BTW, I think a manifold looks and works better than a Y adapter.

I cant tell you wheter manifold works better or not than Y-adapter. However, if you were to use either one, I would suggest putting it right after the CPU and one tube going into GPU.

As for the pump, as far as i know, runing two pump in series (lower pressure feeding higher) will give you the max head.

Lose the res too...it will help you gain more flow. Also, if you havent already bought the pumps, Id combine the money and get a monster pump like an Iwaki. Heres how I would set it up:

pump>WW>Y-spliiter>gpu>rad>pump

just as long as you hit the cpu first with the cool water, everything else should be fine. Otherwise, if you wanted to stick with two pumps, then id go:

pump1>WW>y-spliiter>pump2>gpu>rad>pump1

Lose the manifolds and res, they will clutter your system and give more restriction.

just make a res that directly screws into your pump like i did with my eheim. i get the same temps that i did with my t-line and my system bleeds a lot faster.

cupidr17 said:

... Otherwise, if you wanted to stick with two pumps, then id go:

pump1>WW>y-spliiter>pump2>gpu>rad>pump1...

Click to expand...

Why would you do that? Seems like a lot of hassle, considering that the pump placement doesn't really mean a whole lot. Better to do as ILikeMy240sx says and have the weaker pump feed the stronger pump directly. Unless you are operating at a point that is beyond the P-Q curve of one pump (highly unlikely) then the two pumps will work together in series as one without a whole lot of troubles.

Running them in parallel will only hurt their performance... pond and aquarium pumps are already quite good at high flow - low head operation, so there's no reason to help them out eh.

It looks to me like your setup will cause imbalance, the side of the WW going to gpu will have less flow. Not sure how ths will effect the performance of the block.
I see 2 options:
1) get rid of the manifolds and res, add a small rad for gpu on run them as seperate loops.
2) semi-seperate loops with shared rad & res. have both blocks goto a manifold and then into the rad, which then goes to res, where each pump sends to one block.

Hope those make sense, I don't have PS installed yet, just formatted everything. If I can find it later, maybe I'll pot pics if ya don't understand what I'm trying to say.

peace.
unloaded

A few problems. There is no "weaker" pump, they are both the exact same pump. I have been running them in series on my old machine and they seem to work fine, but I wanted to mabey try something better. Lift is not important since the max drop will be under 12". I am trying to get MAX pressure out of both pumps. And they are Via1300 so I use both of them in case one of them dies.

Sorry, there is no way I am losing my res. I spent to much time making it. You can see it
HERE .


UnLoadeD:
If I get it I like option number two, but I am a big fan of going Pump>Radiator>CPU>GPU>RES.....

But... If the WW does not flow all that well anyway mabey it would not really matter that much?

One last thing. One reason I wanted to use a manifold for the pumps is that they are the small 3/8" fittings and would be VERY hard to dig out and replace. I figured I would not lose as much if I used a manifold to adapt from 3/8" to 1/2". I may be wrong though. Also I am in kind of a bind, you can see that I made my res with two outlets thinking I knew how it was going to work.....

skahtul said:

A few problems. There is no "weaker" pump, they are both the exact same pump. I have been running them in series on my old machine and they seem to work fine, but I wanted to mabey try something better. Lift is not important since the max drop will be under 12". I am trying to get MAX pressure out of both pumps. And they are Via1300 so I use both of them in case one of them dies.

Click to expand...

That's great news... this means that it doesn't matter which is pump feeds the other. Your comment about the lift only being 12" shows a lack of understanding of how the pump will interact with the system... in fact, the lift is mainly irrelevant. What the pumps operate against is the resistance caused by the blocks and fittings... the aim is not to maximize the pressure, but to maximize flow by reducing the pressure drop through the system. Indeed it will be the series pump arrangement that will lead to the best system flow.

P.S. If you seriously want the MAX pressure out of both pumps, just cap the ends.

But am I correct in that I can achieve better/more flow with two pumps or is that not correct?

Can't Sleep....

I can't sleep because I'm sick tonite, so I'll chime in here....

First, the WW block performs almost as well using one outlet only vs. both outlets, so do not be concerned too much about much more flow exiting one outlet vs the other.

And my ww block is far less restrictive than any other CPU block I have used, for what it is worth.

I have three blocks in my system, so I ran the "high flow" exit from my WW to my GPU and the "lower flow" exit to my NB block. Both those blocks have smaller 3/8 ID, but because the flow rate to each is half of what is going into the WW, they really are not a contriction becasue there are two 3/8 pathways now vs. one 1/2.

My original statement:

Pumps in series also will get you higher flow than a single pump, but not as much as two in parallel (given your system is not too constrictive).

EDIT: I got this backwards, sorry, two pumps in series will yield BETTER flow than two in parallel. Thanks Skulemate for catching this error!

As you have already invested quite a bit in your project, I'd get a Northbridge block and run it in parallel with the GPU block directly from the WW, which turns the WW into your manifold...(sweet!)

As far as what order?

Start with the radiator out to the WW in.

If you have only the GPU block left to cool, take the "high flow" WW exit and run it by the GPU, and the GPU exit to inlet of pump 1.

Run your "lower flow" ww exit straight to inlet of pump 2 (insert northbridge block here before the pump if you decide to get one) . Take the outputs of both pumps and go to the Resivoir (which is pretty, but an unnescessary restriction IMHO).

Then from Res to Rad.

If you don't have enough flow through the GPU, artificially constrict the other parallel pathway to force more water by the GPU. You can use wire ties to clamp it a little.

I've always ran mine pump->cpu->rad mostly because it allowed least amount of tubing and turns. I never had a problem like that. I also have a WW block. I have one made by cathar, copper top and BSP threads on the barbs. I did a bit of math and found that 2x 3/8" ID has almost the exact size as 1x 1/2" ID. My plan was to use the 1/2" for inlet and 3/8's for outlets, then use one of those 2x3/8 to 1x1/2 adaptor Y's. The reason I wanted to do it was to gain a bit of elbow room on the block itself. With thick walled 1/2" tubing on all 3 barbs, I didn't have enough room for hose clamps. The trouble I ran into was finding any BSP threaded barbs to replace the ones on my block. If you have the Dtek block I'm sure it has the NPT threads. You could switch the outlets to 3/8". Then add a NB block. Run one outlet to gpu, other to NB. Then bring the gpu & NB outlets together with an adaptor Y and goto your res. This should create better balance in the WW and it would also eliminate the manifolds. Also pressure drop shouldn't be a problem if you are running both pumps inline. I'll try to find you a link to those adaptor Y's, I thought Dtek had them, but I don't see them. I haven't transferred my bookmarks over yet, so if anybody else can link to them, please do.

peace.
unloaded

edit: hmm, seems me and zippy have similar ideas. I don't know about the "high flow" and "lower flow" exits on the WW block. They sould both be the same unless there is a restriction further down the line.

Clarification....

The WW block's center inlet is actually slightly off-center relative to the channels (remember the 3 dots on one end?). If you look carefully down the center inlet you will see the aperatures of the channels are aligned so that more water will naurally flow one direction VS. the other. This is what I ment by high VS. low flow.

There is not much of a difference, but there IS a difference.

Rather than introduce unnescessary reducers to the mix (or change fitting size on the WW), I simply used 3/8" id hose for the two WW exits (when heated in water and given lubrication, 3/8 ID hose will fit on the 1/2 ID WW barbs).

One went to GPU in a gentle bend, and for the other WW exit I deliberately used two pretty constrictive 1/2 ID 90's to make a "U"/short run to the Northy. (I used thick walled 1/2" ID tubing and a pair of oppossing direction wire ties on each WW connection. Fit fine as I had the thinner 3/8 Hose walls on the outside barbs.)

This resulted in much higher flow to the GPU and lower flow to the Northy....this was my desired optimization as the Northy does not benefit nearly as much from more vs. less water compared to the GPU. I'd "guess" that 66 Pct of the flow goes by the GPU, based on bubble speed during priming.

So I have only 1 "Y" in the system (1/2" ID on all arms). I force fit the two 3/8 hoses exiting the GPU and CPU on the y, and used 1/2 ID to return to the pump. Voila....

Re: Can't Sleep....

zippyc said:

... I actually like the idea of using two pumps in parallel for higher flowrates. Pumps in series also will get you higher flow than a single pump, but not as much as two in parallel (given your system is not too constrictive).
...

Click to expand...

Well then, it's too bad that the idea is faulty in the first place. You will not get higher flow rates with pumps in parallel than you would in series... sure, parallel pumps will try to double the flow for a given resistance, but that will only cause the resistance to increase to match the new flow... which is something that the pumps cannot handle. Remember, the system resistance will vary proportionally to Q^2. What will happen is each pump will end up delivering less flow than it would alone in the setup, meaning more heat will be transferred to the water, and the pump will be forced to struggle against a higher head than it should have to lowering its efficiency. If you want to try, go ahead, but you are wasting time.

Re: Re: Can't Sleep....

Skulemate said:

Well then, it's too bad that the idea is faulty in the first place. You will not get higher flow rates with pumps in parallel than you would in series... sure, parallel pumps will try to double the flow for a given resistance, but that will only cause the resistance to increase to match the new flow... which is something that the pumps cannot handle. Remember, the system resistance will vary proportionally to Q^2. What will happen is each pump will end up delivering less flow than it would alone in the setup, meaning more heat will be transferred to the water, and the pump will be forced to struggle against a higher head than it should have to lowering its efficiency. If you want to try, go ahead, but you are wasting time.

Click to expand...

I can see where that makes sense. What is it that will happen when the pumps are in series?

Glad that some other people here have the WW's and have chimed in. For now there will not be any NB cooler. I am getting a new video card for Christmas and the GPU cooler on my TI200 will be used for a NB cooler then. But then again, I may not be done until then so maybe there will be a NB cooler to start with.

The more I think about it the more I want to put in a third pump and put my GPU on a separate loop with no RES. The radiator I have for my CPU is not huge and I have a small 4.5" Square radiator left over from another project that I could use for the GPU and NB. If my CPU were all by it's self I would have less variables and be able to tweak it a lot easier.

Why does a res. restrict so much? I have always used one and am just wondering.


BTW, In case anyone wants to see my project (so far) you can see it HERE

Nice case, you definitely have room for 2 seperate loops. That heater core from your old project will cool better than both of those tranny coolers combined, also its much less restrictive than either of them. Another thing to consider is that aluminum will corrode pretty easily in a loop containing copper. If you use them, be sure to research corrosion inhibitors and the correct ratio they should be used.

Why are you dead set on using 2 pumps for your cpu loop? I have a Via/heatercore/t-line for my WW loop, it does great. If you are that worried about pumps dying, maybe you should get something you feel more comfortable with?

peace.
unloaded

UnLoadeD said:

Nice case, you definitely have room for 2 seperate loops. That heater core from your old project will cool better than both of those tranny coolers combined, also its much less restrictive than either of them. Another thing to consider is that aluminum will corrode pretty easily in a loop containing copper. If you use them, be sure to research corrosion inhibitors and the correct ratio they should be used.

Why are you dead set on using 2 pumps for your cpu loop? I have a Via/heatercore/t-line for my WW loop, it does great. If you are that worried about pumps dying, maybe you should get something you feel more comfortable with?

peace.
unloaded

Click to expand...

Yeah, corosion. I have had that other setup for over a year and never had a problem, it mixes copper and aluminum. I just put a little additive in and haven't had any problems.
Dual pumps for more/better flow is the main reason for dulies. I don't have the money right now to get something bigger or better.
The heater core is really good at cooling and has good flow but it is for my refrigeration project. I used that tiny oil cooler buy it's self on a nother project to cool my CPU and it actually did pretty good. That heater core is huge, it is to big for my case. There are 4 92mm fans on top of it and that does not even cover it all the way.

Skulemate, you are correct...

I suppose my memory is not at it's best at 2 AM with a 101F fever!

I went back to my tests and using the same lengths of hose and same block, it takes two pumps in parallel 2 min 26 Sec to lower the level in a full five gal bucket down to the 1 gal mark I taped to the side of the bucket.

Same two pumps in series with the same block took right at 1Minutes 59 Seconds.

So the system took a 15 Percent flowrate hit in parallel vs series.

I suppose this is why I chose to use two pumps in series (one push one pull) to pump though my pair of heat exchangers which have two large cores in series (each, so I was using four pumps total.)

The pumps I used were the Maxijet MP1200's.


Good catch!