Pump testing

[pHaestus]

Ok guys here's the deal. Knightelite is coming over to the house tomorrow to help me with doing a bit of testing. We settled upon an interesting topic brought up by BillA recently: REAL performance of pumps in a watercooling loop. So there are two ways to do that sort of testing. 1) Just hook a "typical loop" and a flowmeter to a bunch of different pumps and measure the resulting flow rates. 2) Add a valve and a pressure gauge as well. Measure pressure drop across the pump over a variety of flow rates, and construct a P-Q curve. You have to account for the head loss due to the flowmeter and valve so those will need to be measured as well. This is how I was planning to set things up for the second sort of testing (suggested by Bill in a previous thread):

http://phaestus.procooling.com/pumptest.jpg

This is where you guys come in. I have two questions regarding the impending flurry of test activity.

1) What constitutes a "typical" setup? Are more than one setup needed? One could, for example test with 3/8" tubing and 1/2" tubing to create a "high restriction" vs. "High flow" scenario. The parts must be identical for all tests of course. I was thinking that maybe a Black Ice Extreme or a Chevette heatercore (same as DD and DTek use) and a CPU waterblock would be the norm. Should I add a GPU block too? I have a few CPU blocks available but the Innovatek GPU block is the only one I have. I could say add 2 CPU blocks and a GPU block and a radiator with 3/8" tubing for a "worst case" type scenario too. Seem reasonable?

2) Are you guys going to find the P-Q curves useful? I would guess that the bulk of the pumps will fall right along the mfgr specs and so it will be an instance of lots of effort for something we already knew. There is some indication that the Eheim numbers might be a bit optimistic though, and several common pumps have P-Q curves created while submerged using a 1/2" NPT intake hole even though they are used inline with 1/2" barbs by our community. Does the final product (P-Q curves from my bench) justify the effort (this always takes a lot longer than you'd think)?

So anyway give me any comments on choice of parts for the "real world" testing and give me your interest level in the P-Q curve generation and when I wake up tomorrow I'll get started. I'll do the real world flow rate testing stuff first so I can go ahead and post that in any case.

[BrianW]

Definitely add more than one block, ideally 3. Also test different configs. Paralel vs serial, even paralel vs serial with rads. Also test with both the extreme and a pro core, maybe even a larger heater core as well.

BrianW

[pHaestus]

That's getting me to the core of the problem with "system testing" pretty quick. I can run a couple of configurations per pump, but not 9-10 for every one. I can take one pump and sort of run a range of blocks and rads on it though I guess. I was planing on testing

Hydor L20
Hydor L30
Eheim 1048
Eheim 1250
Hydrothruster 500 (Vrooom)
Laguna Statuary model3

all at once and then run any future pumps (Swiftech is sending me their new kit for testing for example) as they arrive. I was thinking one page per pump with dimensions, noise/vibration info, and testing results. Just add a new page as new pumps are tested and hopefully we'll have a decent set of data before too long.

[Since87]

1. I think the most restrictive loop would provide the most interesting data. (for me anyway) My impression is that including GPU blocks in the loop is becoming the norm. I'd suggest sticking with 1/2" tubing though. Using 3/8" makes bend radiuses more critical in determining restriction and I presume you won't be able to exactly match the tube routing from test to test, just because of variations in pump size. I'd like to see data for either:

One block as restrictive as a White Water + your GPU block

Or, two blocks that are less restrictive than a White Water + your GPU block.

2. I'm always interested in more data, so yeah I'd like to see the pump curves you come up with, but I'll take what I can get.

[Joe]

what about something like a HWLab radiator.. they are nicely constrictive in some cases.

edit Ahh missed the BI Extreme part.

What about the lil 80mm versions?

[pHaestus]

We are limited, more or less, to what I have here at the house. That includes:

CPU blocks:
2x Swiftech MCW462-UH
Maze 3
Maze 4
2x Spir@l (one's alum and one's copper)
um I think I have a couple others around here perhaps

GPU blocks:
Innovatek with 1/2" barbs

Radiators:
Black Ice Extreme
DD Super Cube (so many bends)
Chevette heatercore
1970 Blazer heatercore (10x6x2 one pass)
BeCooling 5x10 (tranny cooler)

I can plumb in whatever you guys like, but doing all possible combinations of those blocks, rads, and pumps is not a good use of my time. So should I go with a high flow/low resistance type setup (maybe the one pass core and a spir@l or MCW and little else), and a high resistance one (2 CPU blocks, a GPU block, and the Super Cube or maybe a couple of rads)? I am a bit concerned about using radiators that noone actually uses any more though.

Since87 would the P-Q curves be useful for your computer simulation program? If so then I need to make sure I collect the data in a way relevant to you.

[Les]

My interest would lie in:-
1) The P/Q curves for Hydor L20, Hydor L30, Eheim 1048, and Eheim 1250 . A comparison to be posted in your thread http://forums.procooling.com/vbb/sho...?threadid=5068
2) Data for the Swiftech MCW462-UH. For some attempt at cross-correlation with Bill's data - http://thermal-management-testing.co...st_results.htm

[BillA]

be interesting to see the correlation with these tests
http://www.swiftnets.com/Technical/R...mp-testing.asp

I do believe you should test both 3/8" and 1/2" ID

now go to work

[Since87]

Originally posted by pHaestus

Since87 would the P-Q curves be useful for your computer simulation program? If so then I need to make sure I collect the data in a way relevant to you.

Well, with the loss of gmat as lead programmer and the recent dearth of new component data, I haven't considered it likely that the simulator would ever get developed but...

I guess any series of P-Q points is of relevance, my preference would be that units are mH2O and lpm, but conversion is not a problem. In doing the testing I'd suggest recording the data in whatever units your instruments provide, and not trying to do conversions on the fly. (Less chance of introducing an error.) Conversions can be done on all the data at once with a spreadsheet.

I'm certainly interested in whether your tests match the manufacturer's specs.

In addition to the pump data, I'd be interested in knowing the P-Q data for a lot of the components you have. Particularly:

Black Ice Extreme (Very popular and no data)
Chevette heatercore (Provides a cross check between your setup and Bill's)
1970 Blazer heatercore (provides an additional dataset to check against my heatercore spreadsheet)

Maze 3 & 4 (both popular but no P-Q data available)

Innovatek GPU block (I've never seen any P-Q data for any GPU block)

I know most of this is outside the scope of what you are testing, I'm just listing things I'm interested in. If you run out of things to test before allotted testing time elapses...

I think a good idea for future testing would be a GPU block roundup. Lots of people are tossing GPU blocks into their systems (myself included - MCW50) without knowing how the GPU block is impacting the system.

[Blackeagle]

First,

Welcome back BillA! ! Glad to see you back.

Second,

If we wish to see a wide range of testing done, WE need to step up and provide some of the items for pH to test. It's unreasonable to post a list of what we want to see, while posting no offers of providing ANYTHING to aid in a good comprehensive test.

Third,

pHaestus,

Would you be interested in including in your testing any/all of the following?

1) Iwaki MD-15 = max free flow of 300 GPH @ 11.1 ft. of head. Perhaps someone else has a MD-20Z to loan for testing?

2) Rad #2-342 ( the same one Bill did some testing on.) This rad might be usefull in a test for highest system flow rate as it's really low resistence.

3. Cascade (I don't have a White Water, but this would be a high restriction block to use).

pHaestus I think your idea of testing of pump flows through various system configurations can, in time be just as valuable as the heater core data base has been to so many. A fine idea and effort. And like you already touched on, many (most?) water cooling systems have barbs in them, which add a good deal of restriction.

My one added suggestion is to add JUST ONE 90 fitting to show all who read this data in the future how damaging to flow they are. Perhaps adding the 90 into the highest flowing & lowest flowing configurations, but not any others, to save time.

Shoot me a PM if you're interested in testing any/all the above components.

[Blackeagle]

Off topic,

Just what did happen to/with gmat? He along with Morphling1 are very much missed. They both offered so much to the sites threads.

[pHaestus]

Yes I too miss having morphling and gmat around. Hopefully they've been summering at some beach resort with plenty of topless beaches and will return soon.

As far as my borrowing items for testing goes, let me get the test setup in a stable spot and start generating numbers before I borrow anything. Thanks for the offer though and I'll take you up on it sometime (hopefully soon).

Just cleaned up my work area enough that I can get started after breakfast.

[Blackeagle]

Originally posted by pHaestus
Yes I too miss having morphling and gmat around. Hopefully they've been summering at some beach resort with plenty of topless beaches.........

If they have found such a place, they may never be heard from again!

Let me know when you're ready for those components. Be a bit yet before I have them in a running system, unfortunatly.

[pHaestus]

Well at some point they'll have to return home to upload pics + divx. Because we all know if you don't post pics + divx of something like that then it didn't actually happen

Ok I settled upon my first configuration:

1/2" tubing
DD Maze 4
Black Ice Extreme

I used wood screws and zip ties to make it so my flowmeter and waterblock are solidly held down on the table. I also used duct tape to mark the position of the radiator and res, so hopefully hose positioning for the loop can be pretty much identical for all of these. I just finished testing the L20 and am going to put the 1048 on next. No sign of KE still and he has the 1250

[BillA]

seems the long way 'round the maypole
why not generate a dp vs. flow rate curve for the rad, then put the pressure taps on the rad
this will give you a flow rate w/o the head loss due to the flow meter
(which has its own curve as well)

am I missing something ?

[pHaestus]

The Omega is a bit fussy, though I did just plumb it into the loop with Chevette heatercore and I am now taking deltaP measurements. It fluctuates though (as I am sure you recall) I am guessing with the pump's cycling? So I am looking at the inch H2O reading for the outlet of the radiator fluctuate; won't that make it a bit of a challenge to accurately assign flow rates from the deltaP?

It's not trivial to write them down as fast as they fluctuate even...

[BillA]

I used to stare at it hypnotically, repeating the hi-lo values 'till I had confidence in the range
- later I'd calc the mean

the variation will no doubt be different for each pump, bummer

[pHaestus]

Ok I tried to get dP/dQ numbers but the Omega pressure gauge is going to take a little more messing with before I feel really confident with it. So as usual not as much as I'd like was done today.

As I mentioned above, the first configuration would be considered a fairly typical one: Black Ice Extreme, Dangerden Maze4, 1/2" Clearflex tubing. When plumbed into a loop with my flowmeter and a 1gallon res, with stock barbs on the hydor and the ehime, with a 3/8" copper pipe as an inlet on the Laguna (it was modified from submerged to inline) and with a 5/8" hosebarb on the Hydrothruster's inlet, the flow rates were:

Hydor L20: 1.27 GPM
Eheim 1048: 1.32 GPM
Laguna Model 3: 1.59 GPM
Hydrothruster 500: 3.27GPM

These numbers surprise me a little because from the MFGR specs:

http://www.procooling.com/users/phaestus/pqcurves.jpg

I would have expected the L20 to have a higher flow rate than the 1048.

KE never came over so I don't have his Eheim 1250, and the Hydor L30 is being repaired right now after I ripped my pump/res off the inlet. It should all be dry now so I can take the clamps off tomorrow and test it out as well (and maybe get KE's Eheim).

The hydrothruster is in no way suitable for normal watercooling use. It's quite simply loud as hell at full blast.

I'll keep playing with my pressure gauge this week and see if I can't get the L30 and 1250 numbers for the above system and then switch to a bigger heatercore. 2 wbs, and a GPU block for a "high restriction" case as well.

[pHaestus]

Bill are you saying that you just watch the numbers for as long as it takes to feel comfortable that you have seen the max and minimum value for the high side, switch the valves, and do the same for the low side? Then use the means of high and low to calculate delta P?

[Les]

Originally posted by pHaestus


Hydor L20: 1.27 GPM
Eheim 1048: 1.32 GPM
......

These numbers surprise me a little because from the MFGR specs:

http://www.procooling.com/users/phaestus/pqcurves.jpg

I would have expected the L20 to have a higher flow rate than the 1048.
.......

Oh?
It is not how I read the P/Q curves.
To me your results seem in line with the mfg's curves

[pHaestus]

Wow you are right I have a brain palpatation there and was reading 1.27GPM as 127GPH. Sorry about that (time for bed)

The pressure gauge is going to be my bane for a few evenings; I can just sense it.

[BillA]

Originally posted by pHaestus
Bill are you saying that you just watch the numbers for as long as it takes to feel comfortable that you have seen the max and minimum value for the high side, switch the valves, and do the same for the low side? Then use the means of high and low to calculate delta P?

yup, a pain in that it requires total concentration
blank off the last digit (as I recall it has 'too many')

[pHaestus]

Hydor L30: 1.65GPM

It was doing a lot of rattling though; perhaps its impeller needs a replacement. It's possible it was slightly damaged in the res-pump conversion; I may see about scrounging up another L30 from somewhere.

Not an awful performance from this pump (it's in line with the Laguna which is also ~350 GPH at 0 head).

One does not get a lot more performance from the L30 compared to the L20, though. Especially when one considers the price and size differences.

[pHaestus]

I have secured the Eheim 1250 and will plumb it into my loop tonight. I am also planning on running by a few hardware stores to look for "Y" adapters for some more interesting tests.

[Joe]

hehehe Yeh the Hydrothruster is a loud bitch, but man can it move the water

also it generates a shitload of heat