2xD5 vs 2xDDC long loop comparison (serial scaling)

[Martinm210]

Actually, you can calculate and estimate it fairly accurately in all scenarios. You can probably find an old flow rate estimator speadsheet of mine I made a while back.

I had a spreadsheet developed that would add up the restriction curves, add up the pump curves and solve for flow rate. Unfortunately its pretty outdated anymore.

I became interested from questions like this post and wanted to understand it better. Bottom line is that pump performance varies depending on flow rate. The only way to capture the full performance of a pump is to compare the entire curve and its very possible to have two different pumps perform better each in their own respective areas.

Reality is though that even if you can measure a flow rate difference, it doesn't mean much in terms of temperature...and it means even less in terms of overclock.

[luke997]

Actually, you can calculate and estimate it fairly accurately in all scenarios. You can probably find an old flow rate estimator speadsheet of mine I made a while back.

I had a spreadsheet developed that would add up the restriction curves, add up the pump curves and solve for flow rate. Unfortunately its pretty outdated anymore.

I became interested from questions like this post and wanted to understand it better. Bottom line is that pump performance varies depending on flow rate. The only way to capture the full performance of a pump is to compare the entire curve and its very possible to have two different pumps perform better each in their own respective areas.

Reality is though that even if you can measure a flow rate difference, it doesn't mean much in terms of temperature...and it means even less in terms of overclock.

This is actually a problem

My first step while planning the loop and choosing components was using your (btw thanks for that!) and Andreas spreadsheet to estimate the flow.

In both cases I got ~1.6-1.8 gpm - hence I wasn't happy getting ~0.8 and wanted to do something about it.

Anyways thanks for looking into this!

[NaeKuh]

Yes, but 3 of those 7 are acting as 1 block so your loop actually thinks it has 4. Had you stayed serial and with your fancy elbow setup, the DDC's would be much better. You basically streamlined your loop to the point were it doesn't matter which pump you use now.

Funny thought (and I'm not making you or asking you to do it, just putting it out there), I wonder what the DDC's would do if you put them in parallel.

WL answered it for me.

Waterlogged: but without closing one of subloops in parallel, won't most of flow go skipping resistive-lot-of-waterblocks-loop part, when one of pumps fails?

Which is why i love my DDC top.

2 inlets for paralell and then they come together @ the push in serial.

[luke997]

WL answered it for me.

Not trying to argue here but I'm not convinced really - 200l/h (0.88gpm) does not tell me it is high flow streamlined loop - it tells me it is high restriction loop and pair of DDCs in series in such situation doesn't work better than pair of D5s.

That was pretty much whole point of the post - to present some interesting findings for my loop - which is contrary to the usual "DDCs are better than D5s in high restriction loop" and as such might be useful to others (or at least people with know-how and testing benches can test it more and see what's the real story is).

[NaeKuh]

Not trying to argue here but I'm not convinced really - 200l/h (0.88gpm) does not tell me it is high flow streamlined loop - it tells me it is high restriction loop and pair of DDCs in series in such situation doesn't work better than pair of D5s.

well the curve charts say otherwise.

And i was under the impression you didnt have as much stuff on the loop.
Your post says u reduced restriction from what i saw.

Martin even says otherwise.
I have other sources that says otherwise.

Also you changed EQ on both tests didnt you?
Did you keep your platform same for both tests?

Your results and curve charts are contradicting each other.

The only way to capture the full performance of a pump is to compare the entire curve and its very possible to have two different pumps perform better each in their own respective areas.

^ see... in a restrictive loop, a typical closed loop in our hobby, the DDC is king.
Add restrictions more.. ie.. more length in tubing... restrictive blocks, the DDC only pushes further ahead.

You have a low restriction loop, with a PA and a Niagra, then i would see the D5 winning.
But less then 1% of our hobbiest will be on a setup like that.

[Church]

Imho waterlogged's choice of words make it sound a little bit worse then it is. It's not that paralelising 3 GPUs made DDCs loose their performance, it's just that by lessening overall restriction this way D5s gained some. Of course, relative difference would be higher with connecting GPUs back to serial, and flow would be less with any pumps, just that D5 will see more of a drop.
Still, 0.88gpm seems a bit too low for that much pumping power, for both dual DDCs & dual D5s. I'd expect at least 1.5gpm with these blocks/these pumps. I wonder if there isn't something else wrong with loop. Are pumps not undervolted for shure? No kinks in tubing? Are cpu waterblock inlet&outlet connected as they should/not mixed up? It isn't the case with some fittings having too long of a thread and screwed in too deep in waterblocks/full cover blocks to humper flow? How many 90deg fittings used? I would test each of loop parts one by one to find abnormal flow resistance culprit.

[Waterlogged]

Thanks for suggestions and diagram.
Changes look actually much easier than I though there would be (no need for another res, etc.), so I actually might pick up some fittings and single tops in the next couple weeks out of curiosity.

That's pretty much the order I have - so basically what you're saying is that there is not enough restriction in the loop for the serial setup to benefit it and the parallel setup would do the trick and improve the flow significantly?

Also, considering what the current values and scaling in series are any ideas what numbers for the parallel might be?

For example currently:
1 D5 = 130,
1 DDC = 160

2 D5 series = 200
2 DDC series = 200 (I've adjusted from 205 for GPU serial to parallel change)

2 D5 parallel = X?
2 DDC parallel = Y?

I realise it's hard to tell with that many variables but perhaps interpolating from some existing tests I could have some idea?

P.S. Sorry for double post it - didn't notice this post before replying earlier.

Your guess is as good as mine Luke.

I'm busy trying to test out Churchy's point to see just how much of a concern it should be but, I need to go to work now and I won't be able to work on it for at least 8hrs. I'll try and get something done when I get home but can't guarantee anything, may take a couple more days to see any results for this as I've got a couple planned posts full of pics already in the works that I'd like to get rid of first.

While your waiting, could you check the flow meter to make sure it's setup properly? Something sounds really off if there's that much difference from Martin's (or even Andrea's) calculator.

[NaeKuh]

wait im rereading his post again.

Did he say he got more flow on a D5 with more restirction then a DDC?

Because im very lost now in his post.

[luke997]

wait im rereading his post again.

Did he say he got more flow on a D5 with more restirction then a DDC?

Because im very lost now in his post.

I got 200l/h with 2xD5 with setup "A" for simplicity.

Reducing restriction (by changing rad and serial GPU to parallel) and swapping pumps to 2xDDC got me 205l/h.

If I attribute 5l/h to rad/GPU change, that makes D5 and DDC flow same - if more, than D5s are better.

Again, 1 DDC vs 1 D5 wins in same situation (160l/h vs 130l/h) hence my conclusion of 2x D5 scaling better.

Is it any more clear now?

Again I've said it's no scientific testing here, just an observation.


I have to say I'm even more confused now with you're earlier post about curves - also you're saying my loop is low restriction - but if it is so I should be getting much higher flow, shouldn't I?

Your guess is as good as mine Luke.

I'm busy trying to test out Churchy's point to see just how much of a concern it should be but, I need to go to work now and I won't be able to work on it for at least 8hrs. I'll try and get something done when I get home but can't guarantee anything, may take a couple more days to see any results for this as I've got a couple planned posts full of pics already in the works that I'd like to get rid of first.

While your waiting, could you check the flow meter to make sure it's setup properly? Something sounds really off if there's that much difference from Martin's (or even Andrea's) calculator.

Thanks again WL!

Flow meter can be setup in anyway and the setting on aquacomputer is correct (169imp) - but if it is damaged and showing low flow - no idea

I guess it would make a noise or give no reading in such situation?

[Church]

luke997: most precise way of testing flow (without any flowmeters and without need to calibrate them) - to disconnect loop before and after res, put tubing before pump into large bucket with water and other end in some other bucket (or use bathtub as one of them ). Then time how fast will some specific ammount of water be pumped through. This way you can check if your flowmeter doesn't lie and is calibrated properly.

[Waterlogged]

OK, so with 1 of those previously mentioned posts out of the way, I pushed this up and have something to report.

Yes, it will flow back through the pump into the res but, I think it can be limited as to how much with the right fittings. I was using a DD "T" for the junction where the pumps went serial and that gives the flow a straight path to the other pump should it fail. If a Y is used instead, the flow will be less inclined to make the U turn needed to take that path. It'll still probably happen, but to a much lesser degree. Even with the setup I had (similar to the diagram I made), short of fully closing the valve on my testing res, water was still coming out the proper fitting. Personally, if I were you, I'd give it a try but, that's because I love to tinker too much.

As for the flow meter, are you 100% positive that it is calibrated properly (actually pumping 200L/H)?

[luke997]

As for the flow meter, are you 100% positive that it is calibrated properly (actually pumping 200L/H)?

Yes, I have checked this few times, 169imp is correct value for HighFlow sensor with Aquaero.

Like Churchy though said until checked with bucket or other sensor I can never be 100% sure if there is everything OK with the sensor itself

I have to say though, after benching session with the RX360 and GPU parallel my delta is much better now and GPU temps are more even (which suprised me).

[Serpentarius]

http://www.engineeringtoolbox.com/pu...ial-d_636.html


adding another reference
http://www.rainforrent.com/Training/PumpTraining.pdf

[Church]

I have to say though, after benching session with the RX360 and GPU parallel my delta is much better now and GPU temps are more even (which suprised me).

Maybe because your loop wasn't fully bled & with air trapped in few places?

[Alexandr0s]

OK, so with 1 of those previously mentioned posts out of the way, I pushed this up and have something to report.

Yes, it will flow back through the pump into the res but, I think it can be limited as to how much with the right fittings. I was using a DD "T" for the junction where the pumps went serial and that gives the flow a straight path to the other pump should it fail. If a Y is used instead, the flow will be less inclined to make the U turn needed to take that path. It'll still probably happen, but to a much lesser degree. Even with the setup I had (similar to the diagram I made), short of fully closing the valve on my testing res, water was still coming out the proper fitting. Personally, if I were you, I'd give it a try but, that's because I love to tinker too much.

As for the flow meter, are you 100% positive that it is calibrated properly (actually pumping 200L/H)?

If you use 2 Y's on both sides, won't some flow be forced into the other pump as well, creating some pressure preventing water at the second Y to flow into the other pump?

[Waterlogged]

If you use 2 Y's on both sides, won't some flow be forced into the other pump as well, creating some pressure preventing water at the second Y to flow into the other pump?

Not sure I'm following you. Are you saying use a Y before and after the pumps?

[Martinm210]

FYI,
Just trying to duplicate your results theoretically using some substitutions for parts and using curves for stand along pumps rather than dual tops.

Dual D5s with older EKv1 tops gives me:


Dual DDC 3.2s with EKv1 tops gives me:


So, theoretically the DDC's should actually edge out the D5s by about .1GPM. Nothing to worry about, but slightly different than what you got at least with this approximation.

Now lets try it with an ultra low restriction loop:

Dual D5s using a DD MC-TDX and one thin triple rad


Now the DDCs


It get's a little closer, but still not quite flopping which pump is better. Perhaps with something even lower in restriction, but it's darn close regardless at really high flow rates. The D5 does simply have a flatter curve and gets closer and closer to the DDC the further right you go.

I suspect the difference you found is actually in the dual tops used. We havn't had much testing on those, so there's might be some oddities we don't really know about there..Not sure.

Anyhow, at least with individual pump tops, I still have a hard time seeing how/when a D5 is going to edge out dual DDC's. Their performance is soo close it's moot, but I wouldn't suggest dual D5s will outperform DDCs in series. Typically it would be opposite that.

I also wouldn't bother with parallel pump setups, we're just generally much more restrictive than that unless you're doing something really odd like parallel for multiple loops, but that get pretty hard to predict and not worth bothering IMHO.

My suggestion, run one pump. Either single DDC or D5 is just fine (close enough to not worry much about the difference) for most loops and two pumps if you want redundancy run in series. I wouldn't bother with parallel setups, we just don't see that low of restriction.

[Alexandr0s]

Not sure I'm following you. Are you saying use a Y before and after the pumps?

Yeah, something like this.

[Waterlogged]

Yeah, something like this.

Ah, I suspected as much so I went ahead and did some more testing when I got home last night. I wrestled up some of my old push in fittings and was able to do 2 different setups with actual Y fittings. Results were both interesting and encouraging.

The first test I did was set up like this.



It showed a decent improvement in a reduced amount of back flow but was still rather obvious it was there by output flow rate and I could feel a decent amount of back flow when I stuck my hand in the res.


The other was test was set up like this.



Again, another marked improvement over the setup above in the output flow rate and because they were both pulling from a single supply line, no flow back into the res. This is not to say that there wasn't flow back though as there was, it was just not getting to the outlet on the res.

This setup does come with a caveat though. You need a larger res>Y tube than Y>pumps and pumps>Y. Basically, you would use a 1/2" I.D. (or double the size of the tube you plan on using for the Y to pumps section) tube from the res>Y and then 1/4" I.D. tubing from the Y to the pumps and then back to the second Y and then back to 1/2" for the rest of the loop. To do the entire setup with the same size tubing would be starving the pumps on the inlet side and hinder max flow rate within the loop.


Personally, if it were my loop and seeing the results I've seen first hand, I'd really give it a try but as a last resort, after I've made certain that the flow meter is properly calibrated and none of the blocks/rads were plugged up with crap.

[luke997]

Personally, if it were my loop and seeing the results I've seen first hand, I'd really give it a try but as a last resort, after I've made certain that the flow meter is properly calibrated and none of the blocks/rads were plugged up with crap.

Thanks again. I might do that in couple weeks.

I've did my homework on XS while planning so I've flushed & flushed and flushed all the parts till the water was crystal clear, shouldn't be any problem there neither.

[Waterlogged]

Pumping clear water and pumping the maximum amount of clear water through parts are very separate things. I would take all the blocks apart and double check for anything abnormal and flow rate test the flow meter (it should read close to a single pump's maximum output) and flow rate test the rads (same method as flow meter). Something is wrong if your flow rate is still that low after all those changes to reduce restriction.

[luke997]

Pumping clear water and pumping the maximum amount of clear water through parts are very separate things. I would take all the blocks apart and double check for anything abnormal and flow rate test the flow meter (it should read close to a single pump's maximum output) and flow rate test the rads (same method as flow meter). Something is wrong if your flow rate is still that low after all those changes to reduce restriction.

Right. I've disassembled blocks so only the rads might be variable.
I'm not so fussed about it now though - especially considering that further troubleshooting will be more time consuming - and as mentioned earlier, getting very good (for such a system of course) delta (~7C with all system @ full bore) so time to bench and play some more.

[luke997]

WL - another interesting thing, when water is warm (benching or playing), flow goes up by 3-6l/h. I had no idea the difference will be that much

[Waterlogged]

WL - another interesting thing, when water is warm (benching or playing), flow goes up by 3-6l/h. I had no idea the difference will be that much

That's because the water expands when it gets warm and builds up pressure in the loop and compresses the air in the res. When this happens, it's more like running a T-line which Martin showed us a while ago always have higher flow rates the loops with res's. If you were to open your res and release that pressure while it's warm, your flow rate would drop again.

[Alexandr0s]

That's because the water expands when it gets warm and builds up pressure in the loop and compresses the air in the res. When this happens, it's more like running a T-line which Martin showed us a while ago always have higher flow rates the loops with res's. If you were to open your res and release that pressure while it's warm, your flow rate would drop again.

If that's the case, how well would this work:



In this case, the res basically functions as a rather large T-line.