I need a device that will split a water-flow into two (or more)
outlets, in such a manner that the flow through each outlet is kept at
(or close to) the same level, despite (minor) differences in
downstream pressure.

The idea is to prevent the water from simply following the "path of
least resistance", thus starving the other outlet(s). Ideally this
should be acomplished with as little flow-resistance as possible. If
feasible, the "device" should have some method for "offsetting" the
flow rates (as in: Some way of regulating the difference in flow rate
between the outlets, so as to allow a manual compensation for
excessive downstream pressure differences).

I believe it should be possible to make such a device without any
moving parts, simply by exploiting various fluid-dynamic principles.
However, not being an expert on this subject, I wouldn't know how to
do proceed, nor what to search the internet for?

When I say "no moving parts", I am of course talking about parts that
would need to move "automatically" (during use) to regulate the flows,
rather than parts that might be moved manually during installation or
later on (ie. If included, the "flow offset" regulator mechanism might
have parts (the regulator-knob, if nothing else) that moves when
manipulated manually, but the mechanism doing the automatic
"balancing" of flows during operation, should not).

The specific application is for a water cooled PC, where I would like
to split the cold water from the radiator into two or more seperate
hoses, going to various cooling-blocks in the PC (CPU, GPU, VRAM,
chipset, disks etc.) Thus the flow-rates will be rather low, and the
pressure even lower (think 5W aquarium pump through 8-10mm silicone
hoses). It is thus important that neither high pressure nor high flow
rates are required for the "device" to work properly.

As you might have guessed, the reason for using such a device, is to
ensure that all parts of the PC recieves (about) equal amounts of
water, preventing all (or most of) the water from going through
whatever part of the hose-system offering the least resistance.
However, since there is typically an excess of cooling-capacity in
such systems, the flow-regulation need not be very precise, and it
would be perfectly acceptable if somewhat less flow were to go through
a path with higher resistance, as long as SOME water is kept flowing
through all paths (any further "tweaking" of how much water goes where
can always be done manually, which is why a regulating mechanism for
this purpose is needed.)

Thanks in advance,

inventus-ga.

---

Clarification of Question by inventus-ga on 11 Jun 2005 22:43 PDT

Hi again,

Just wanted to add this: The difference in "resistance" (pressure?)
through each of the "paths", should be relativly minor, especially
since the pressure is low to begin with.

So if anything, the device should be based on flow-differential,
rather than pressure-differential. Even so, I imagine something using
the ventouri principle or similar, perhaps to reduce the flow through
turbulence-induced resistance, whenever the flows aren't "matched"..?

But as I said, I am no expert :-?


Regards,

inventus-ga

---

Request for Question Clarification by cynthia-ga on 12 Jun 2005 02:08 PDT

Hi inventus,

Perhaps you could use these principles, or simply use this product:

W.A Kates Company
http://www.wakates.com/
..."has been manufacturing flow rate controllers for every type of
application under the sun. From air to zinc hydrosulfate, Kates
controllers provide a simple alternative for numerous complex flow
applications..."

I have no idea what size they are, is this on the right track?  If so,
maybe I can find a DIY thread on a water-cooled PC BBS.

~~Cynthia

---

Clarification of Question by inventus-ga on 12 Jun 2005 17:24 PDT

In response to cynthia-ga (on 12 Jun 2005 02:08 PDT):

Thanks for the linkage Cynthia, but I'm afraid that their products
doesn't seem to meet my requirements (although they DO have some in
what appear to be the right "size range").

First of all their valves are full of moving parts (and even
electronics in some cases), whereas I would like a solution without
such, if possible (as described above). Second, they do not appear to
actually have a product that will "balance" (split?) a flow into two
(or more), while ensuring that the resulting flows are (somewhat)
equal; at least not without using several of their valves and some
electronics, as far as I can surmise?

But thanks anyway, for your time. If you do have linkage to a suitable
BBS thread, and feel it might have any relevance, please don't
hesitate to post it here...

Regards,

inventus-ga.

I'm no expert either, but from your description of the actual use, you
want to assure that both elements are being adequately cooled.
Would it be possible to control your flowsplitter with thermostats
downstream of the elements?  I am assuming existence of some kind of
electronics that would compare the relative temperatures of the water
coming out and then shunt more water to the element with the higher
temperature.  It might even be able to speed up the pump if the
temperatures got too high, or sound an alarm.

Don't ask me if there is something that can do this.  I have great
faith that if I can think of it, someone else has long since done it. 
:-)

Good luck, Myoarin

In response to myoarin-ga (on 13 Jun 2005 06:11 PDT):

Hi again myoarin-ga (I seem to remember you commenting on my question
about, was it cooling below room temp, or some such?)

I have no doubt that one could make a more or less elaborate
electronic setup, using some kind of selenoid valve or similar, and a
(number of) electronic thermometers and/or flow-meters etc.

However, if at all possible, I am after a non-electronic (and
preferably non-mechanical as well) solution to this "flow-adjustment
problem". After all, what is needed is just a limited "regulation", in
case one "path" offers a bit more "resistance" than the other, it is
not like there is any need to regulate with any kind of precision, nor
is there any need to be able to regulate large differences in flow or
pressure.

I am rather sure that I've heard about at least one such device,
capable of (limited) relative flow regulation ("equalizing" two flow
rates), without having ANY movable parts! This, if I remember
correctly, was done purely using a clever arrangement of
interconnected pipes?

I'm guessing it is a matter of exploiting the ventouri effect and/or
generating turbulence at the right place(s), thus making it "harder"
for the water to pass through the outlet with the faster flow of the
two and/or making it easier for the slower one.

The question is if anyone know how this is done, and/or if such a
regulator is currently protected by patents?

But, as always, thanks for your comment. I certainly will look into
the "electronic solution", should nothing useful come of this
question. In that case, the main problem will be finding a suitable,
electrical regulated, valve (all the other stuff: Inline, watertight,
flow- and temp-meters etc. being readily available in models
compatible with existing PC fan/pump managment hardware and software.)


Regards,

inventus-ga.

HI Inventus, yeah, it's me again.  Sorry, I didn't even look at your
name, but I should have recognized the copious explanation.  :-)

I was originally assuming that the flow rate in the two (?) tubes
could vary, I quess with the variance in temperatures, the graphic
card overheating at times, perhaps.  Is this true?  Or would the flow
vary for other reasons?

Or is the system constant, just a greater resistance on one side?

If it is constant, I don't understand the need for a self-regulating
device.  You could just increase the resistance on the other side with
an adjustable screw clamp on the hose, squeezing it until the flows
equalized.
BUT, you'd have thought of that, so I assume the resistance can vary.  Why?

Now that I have searched and learned about the Venturi or Ventouri Principle,
I understand your expectation that a system of tubes back to the valve
could control it, the lower pressure from the fast flowing side,
shifting the valve to reduce the flow to that side.  Sounds like an
elegant solution if it could be made to work.

Interesting, but I am still wondering what varies the flow speed.
Myoarin

(Hmm, once again I seem to have prooven my utter inability to limit my
wordage, apparantly my most notable trait. I hope you can stay awake
for this...)

I guess you are right Myoarin, the pressure or flow difference will
probably be mostly constant (and even if it does vary with
temperature, I'm confident that some "average" setting could be found,
provinding a more than reasonable compromise for the entire operating
range).

But still:

In order to "tune" such "screw clamps" (or any other "flow regulator"
for that matter), I would need some kind of "feedback" (say a
flow-meter in each tube). So no matter how simple you imagine the
"valves", you still need measuring devices for the "feedback" (which
might, of course, also be made "simple"?)

The self-regulating gizmo however, would find some kind of "near
equilibrium", and even if it then remained largely at the same
setting, the fact is that you A) didn't have to "tune" any valves
during the inital setup, and B) won't have to worry about "re-tuning",
should you change some part in one of the water paths (or perhaps just
due to some kind of "drift" over time?)

This also explains why I am looking for a cheap, non-mechanical,
non-electrical, sensor-less, solution. There really is no argument for
any kind of elaborate setup with valves, flowmeters and electronics.

But now you mention it, one might get a long way, merely with a simple
flow-difference (pressure-difference?) meter (might it be as simple as
a ball in a tube connected to each of the measured tubes?) As such a
device would allow any such "clamps" or "valves" to be tuned quickly
(as well as keeping an eye on any change in flow, due temperature,
wear or whatever).

Btw.: The "elegant solution" you describe, with a valve being operated
by the fluids themselves, I am convinced must already exist (if
nothing else, according to your own statement, that some one else must
long since have thought of, and done, this!)

One, poor-mans, design (and a bit alternative at that), simply
involves two pumps (simple "water-mills" might well do the trick),
these simply being directly coupled to eachother, mechanically. Thus
the water would be driven throug each pump at roughly the same
flowrate (or any predefined ratio, if the two pumps vary in volume or
a gear is introduced between them).

The fact is that I want to move a step further than this, and
eliminate the "valve" altogether! I was planning on getting the water
to act as its own "valve", simply by making it "move against itself"
by causing turbulence and counter-flow and/or make it change its path
through the tubes, due changes in flowrate or pressure, or..?

Thus I imagine a device with no moving parts at all!

Of course, this might just turn out to be an impossible pipe-dream (no
pun intended).

Another concept, besides the Ventouri effect and turbulence, that
might be of relevance, is the "injector-pump". Here a small, high
pressure, flow is used to pump a much larger flow (with less
pressure). If some variant of this principle (which might well be
parallel to Ventouri?) could cause (part of) the faster flow to act as
a pump on the slower... (As per the "poor mans" pump-regulator
descibed above, only an "injector-pump" has no moving parts!)


Regards,

inventus-ga.

Two pumps, cheap, sure, but not much fun  (Oh yeah, you were worried
about the noise before, as I remember ....)

IF the flows remain constant, as a test of equal pressure I was just
going to let outlet tubes squirt into a basin and adjust till their
arcs were equivalent (peeing contest).  Then you can connect them
again  -or incorporate them in a little room fountain producing
southing water noise, and have a constant visible control, maybe even
audible control, while also letting the fountain act as an air
humidifier.

That adds a different kind of elegance.  :-)
Myoarin

I think the key is to use the venturi effect.  Have  each flow line
with a flexible side section, and a sprung loaded bit puching against
them.  when flow is high, pressue=low, sprung bit moves in and
restricts the flow. someone clever should be able to work out how to
use this to get a balanced effect.

How about a "y" fitting that splits the flow into two funnels?
Think of it like this: the input stream is split into two reasonbly
equal streams that empty into a right funnel and a left funnel. The
right funnel feeds a section of the cooling system that has little or
no "resistance" or backpressure. The left funnel feeds a portion of
the system that has significantly higher "resistance". The
backpressure on the left side causes the fluid level to rise in the
left funnel.
However, as the fluid level rises, the weight of the fluid in the
funnel causes greater pressure. Eventually, an equlibrium is reached
where the higher pressure on the left side causes the fluid to drain
out of the funnel exactly as fast as it is coming in. Thus, the flow
through the left and right sides is equal, even though the
"resistance" is quite different.

The taller and thinner the funnel is, the more quickly it will reach equillibrium.

If you are indeed using just an aquarium pump, this kind of
arrangement may be quite practical. However higer operating pressures
would require an unwieldly tall funnel arrangement.

Good luck and happy overclocking!

Hi,

Thanks for your comments, maveric99-ga and kc8nod-ga.


maveric99:

Could the "flexible side section" perhaps be shared between the two
pipes, so the lower flow / higher pressure would act as the "sprung
bit", pushing the soft wall into the higher flow / lower pressure
(which would help) flow, thus restricting it?

(I think your idea is good, just looking to simplify it. With the
above, the only "moving part" would be the flexible section, which is
pretty close to my goal.)

... Even better (this is where I was hoping someone with
fluid-dynamics knowledge would step in?) Could the "flexible side
section" be replaced with a (set of?) pipe(s) between the two flows,
such that the mere flow of water through this "interconnection", from
the high pressure / low flow pipe to the low pressure / high flow
pipe, would somehow act in a manner similar to the "flexible side
section", and somehow "disrupt" the high flow?

(This would be ideal, as there would be no moving parts at all, not
even the "flexible side section"!)


kc8nod:

Very clever idea, I hadn't thought about this kind of solution at all
(ie. "gravity assisted"). And it has no moving parts too, just like I
want.

Only problems with this solution, as far as I can see, would be:

A) The "splitter" would need to be mounted "upright" with respect to
gravity (this is not a major problem, just a slight "disadvantage".)

B) If I understand your concept correctly, the "funnels" would need to
be "vented" (ie. there would need to be air over the water surface in
the funnels, which would need to be able to get out/in, as the water
levels rise/fall. If so, a completely blocked or too restricted flow
in either pipe would cause water to exit the system through these "air
vents" (However unlikely it is that this situation should arise, I
don't like the idea of water being able to get out, especially as the
flow-splitter would likely be placed INSIDE the PC!!!)

While A isn't too serious, and thus could be lived with, B would need
to be solved (I'm rather sure a "vent" can be designed so it won't
allow water to escape, not that I know precisely how? A simple, but in
some situations impractical, solution would be to pipe the vents back
to the water reservoir. But it so happens that my reservoir is outside
the PC, which would make it quite bothersome to have a (third) pipe
running out through my case and into my reservoir!)

Overall though, I like the "funnel" part. Perhaps an effect could even
be achieved with non-vented, completely water filled funnels, without
graivty-assist? What if a special "spout" (aperture/..?) injected the
water near the "bottom" (narrow part) of the funnel, and the funnel
had some kind of "secondary outlet" near the "top" (wide part). The
idea being that the water would primarily leave the funnel through the
narrow part (towards which the water was injected), but if/when the
flow becomes restricted (ie. pressure is more "dominant" than flow),
it would also leave through the secondary outlet? (Dunno if this would
work at all though?!?)


I hope both of you (and others) will continue to comment on these
ideas and come up with other suggestions. If/When a sufficiently
promising concept is presented, I will make a prototype and see if I
can get it to work...


Regards,

inventus-ga.