Equal flow distribution in manifold 1

[gkan]

Hi all,

This is probably a concept design problem so I apologize in advance if its trivial.
I need to design a manifold of 100 needles arranged in a 10x10 grid all connected to the bottom of a liquid container.The container itself has 4 inlets on top in which liquid is pumped using a peristaltic pump. The requirement is that each needle will provide with a flow of 10 microliters per second +/- 5 microliters. The needles themselves have an ID of 1 mm. The problem so far is that the flow distribution is not equal among the needles. Some have a flow of 10 microliters, some have no flow at all.
My thought was that by drastically increasing the pressure drop in each inlet of a needle (maybe by introducing a porous sheet with a very small porosity) I could perhaps compensate for the large ID of the needles in comparison to the very low flow rate which seems to create unequal distribution to the 100 outlets. My questions:

1. Has anyone faced such a problem before?
2. Does the porous sheet solution seem viable? if so, does anyone know of a material that could work in this case?

Thank you for reading,

George

 

[TheTick]

I literally grew up with this problem, making melt-blow dies for my dad's company.

The very first thing (if possible) is to have the largest possible plenum to feed the needles. You want to minimize any localized stream flow from plenum intake to exit (needles).

Diffusing the pump discharge into the plenum will also help.

Peristaltic pump seems like a poor choice. You want consistent pressure in the plenum. Some sort of positive displacement pump (a pair of small gear pumps?) with good pressure regulation and maybe even some recirculation to help regulate.

Quality of your needles is of paramount importance. Even microscopic variations in ID and length can cause significant variation.

Temperature regulation counts for a lot if your fluid is viscous. Warmer needles will have more flow than cooler ones.

Honesty may be the best policy, but insanity is a better defense.

http://www.EsoxRepublic.com

-SolidWorks API VB programming help

 

[gkan]

@The Tick

Hello and thank you for the prompt reply. The liquid is not viscous, it is close to 2 cP , so lets say water.
Diffusing the flow was my first thought, but I was afraid it will not be enough. The very low flow rate and the comparably huge size of the needles result in an extremely small pressure drop along the flow paths. This is why I though of introducing a high pressure drop component before the needles, to render the manufacturing deviations among the needles unimportant. Another reason is that I have noticed that even tiny errors in the orientation of the system (not entirely normal to the gravity vector) has a tremendous effect on the flow difference among the needles. The thing is, the setup is pretty much set by the client, so I am trying to make this work with a minimally invasive solution to his device.

 

[JackAction]

I like the concept of your solution, I just would use a rather thick plate of metal with 100 tiny holes, one over each needle. Size the holes such that they control the flow under the maximum pressure your pump can handle. It's like the concept of a carburetor jet:

 

[chicopee]

I would say that if you could feed the plenum from various directions, the flow thru each needle would be more evenly distributed.

 

[MikeHalloran]

It's a nice idea, because they're flow-matched, but carburetor jets are too big.

Maybe a 10x10 array of inserted carriers, each bearing a sapphire orifice with a small hole. You may still have to flow-match them because orifices are so sensitive to aperture diameter and entry and exit radius.

If you don't need continuous flow, maybe a 10x10 array of small plunger pumps, all driven by a translating plate. Then all you need is a 10x10 selector valve to charge the pumps. Easy-peasy; hah!




Mike Halloran
Pembroke Pines, FL, USA

 

[gruntguru]

At the very low flow rates, it is unlikely that velocity distribution in the plenum is a problem (unless the plenum has a ridiculously small cross section). The peristaltic pump is a good choice - positive displacement, accurate metering. Seems your lack of pressure in the plenum (lack of pressure drop across each metering orifice (needle)) is allowing non-uniform factors (gravitational head, capillary action/surface tension etc to dominate)

Yes you need 100 metering points with equal cross section and discharge coefficient, small enough to generate a significant pressure drop - say 50% of the pump rating at the specified flow. The simplest way may be a screen as you suggest. Make sure the 100 holes the screen material rests against are equal. Something like a reverse osmosis membrane might work, the important thing is uniform permeability across its surface. Make sure there are no solids in the working fluid - they will be filtered and block the membrane.

je suis charlie

 

[katmar]

I would go with the ideas put forward by JackAction and GruntGuru - you need an individual restriction upstream of each needle. The head required to get the desired flow through the needles will be a fraction of a mm of head and using the method you have now it will be very difficult to mount all the needles with their top openings at precisely the same level, and to keep the whole arrangement level.

The problem with an upstream restriction is that an orifice with a diameter of 0.2 mm needs a head of only 15 mm or so to give the flow you want. Using an orifice with a diameter smaller than that is risking endless blockages, and even with 0.2 mm you will need to ensure that your fluid is well filtered.

Also, with such a low flow the liquid coming out of each hole is not going to be continuous. It will grow to a droplet size before over coming the surface tension and falling to the needle. To get the drops to fall in exactly the right spot will require some sort of pin or pipe at each orifice that will allow the drop to grow at a specific point and fall where you want it. Rather than trying to get the drops to fall exactly into the needles I would build a grid of "pools" around the needles to catch the drops and allow the liquid to flow into the single needle in each "pool".

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[gruntguru]

As long as all air is bled from the system and the needles are sealed to the plenum, there won't be any dripping between the restrictor and the needle. The drips will form after the needle.

je suis charlie

 

[chicopee]

That statement "At the very low flow rates, it is unlikely that velocity distribution in the plenum is a problem (unless the plenum has a ridiculously small cross section)" by Gruntguru makes a lot of sense, so I agree with it.

 

[rb1957]

also the target isn't equal flows at each nozzle, but 10 +-5 ... anything from 5-15 is on target.

with the largest plenum, maybe add baffle ribs to break up the inlet flows and direct flow into the corners ?

another day in paradise, or is paradise one day closer ?

 

[gruntguru]

rb1957 you still attribute the unequal distribution to dynamic pressure variations in the plenum? At least two posters (chicopee and myself) believe this is unlikely. The OP has not given any indication that the variation in flow has a pattern that would suggest this - e.g. low flow in the corners.

gkan - you could try a sheet of filter paper as a first guess - add layers if the pressure drop is still insufficient.

je suis charlie

 

[gkan]

Hi everyone,

...and thank you for a plethora of ideas! I lean towards the initial idea, which is to use a sort of filter sheet to the inlet of the needles. I think an equal distribution in the plenum would only be useful if dynamic pressure differences were actually the reason for an unevenly distributed flow. In this case even orifices directly below the entry points of the plenum will display low to zero flow if you tilt the plenum for a couple of degrees. The idea of individual choking points is way over the budget of the project plus I really don't want to risk blockage in individual needles. Gruntguru, when you say filter paper, you mean like american coffee filter paper or is there another product?

Thanks again for your time!

Best Regards,

George

 

[gruntguru]

I was thinking of the type found in every chemistry lab, available in many grades, usually circles (which may not suit your plenum) but no doubt in other shapes too. You could try coffee filter paper - I hadn't thought of that - I wouldn't be confident of the uniformity though.

je suis charlie

 

[1gibson]

So what happens when you target 15 microliters max with the configuration as-is? Do you still get 0 out of some, or do you get 5? With the information given, all I see is that it isn't tested thoroughly enough to know for sure that it requires modification.

 

[katmar]

I must have slipped in an extra zero after the decimal when I originally calculated the pressure drop through the needles and came to the conclusion that the head required was "a fraction of a mm". On rechecking and assuming that the needle length is 50 mm I get a head of 2 mm at a flow of 10 microlitres/second. This assumes the fluid is water and neglects any surface tension effects. At 5 and 15 μl/s the heads are 1 and 3 mm respectively. This means that the total variation in the heights of the tops of the needles and the level of the fluid in the container must be around 1 mm. This is tight, but not impossible. The actual heads may be higher than these if surface tension is taken into account, but the allowable variation would probably not change much.

This got me thinking along a different tack - the pressure drops through the needles can be increased by making the neeedles longer. If the needles could be made 100 or 150 mm long the allowable variation could be doubled or tripled. Several posters have pointed out the need to have the levels equal across the whole plenum. If the extra needle length was achieved by having them extended up through the base of the container (with still 50 mm or so projecting below) the 4 entry points could be made below the liquid surface and this would have a stilling effect on the surface level compared with allowing the entering liquid to fall and splash into the liquid.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[gruntguru]

If the needles can be individually moved vertically (e.g. if they were mounted by fitting tightly in o-rings or a perforated rubber sheet) it would be possible to adjust the heights individually to equalise the flows. (Lower needle = more gravitational head = more flow.) katmar's post claims a change of 1mm will produce a flow change of 5 μl/s.

je suis charlie