Air cylinder friction

[johannestoger]

Hi fellow engineers,

I'm a medical imaging engineer way outside my comfort zone, and I hope you can give some insight.

We're trying to generate fast, accurate pulses of water in a 1" pipe. We want a peak velocity of 0.6 m/s and a pulse lenght of 300 ms. Our current idea is to take a 1" diameter air cylinder of the right length (about 100 mm or 4"), and connect an electric servo to the shaft. I found air cylinders with removable end caps, so getting the water in and out shouldn't be a problem.

We're pumping against a low pressure, at most 0.1 bar (1.5 psi) including acceleration, and it's easy to find a servo that can handle that.

However, the friction in the cylinder is a big unknown - what can I expect there? So to give you some concrete questions:

1. What is the typical static friction of the seals in a 1" air cylinder?
2. Can you estimate the dynamic friction when it's moving at 0.6 m/s?
3. What can I expect friction-wise when putting water in the cylinder? The sales reps at a major pneumatics company said "it will probably OK as long as it's stainless steel". The warranty is gone of course. What do you guys think?
4. Do you think I could roll my own using a 1" PVC pipe and a piston sealed with a simple O-ring?

Thanks for reading this far, your input is very welcome!

 

[FACS]

I'll give out my 2 cents.

There is no "Typical" static friction in an air cylinder. It will depend on the manufacture, how many seals, what type of material, lubrication etc... Not only that, but it won't remain constant. As the air cylinder ages, the friction will change. Especially if you are putting water into something that was designed for air.

What you need, is a water pump. We usually design piston pumps for things like this. You can control the stroke for displacement-- it's positive displacement, you can specify what ever seal you want, you can specify the surface finish on the bore ... etc... etc....

And if friction is your concern, then have you looked at the check valves you are planning on using? They have crack pressure and friction loss coefficients; so does the tubing you choose.

But all this is irrelevant, from what I understand so far. The only thing you are trying to control is the velocity and the length ? What is 300 ms? Milli seconds? That's time, not length. Are you looking for .6 meters per second for 300 Milli seconds? Isn't .6 meters in .3 seconds going to give you 180 mm? And the cylinder is 100mm?

Anyhow, what does the friction in the air cylinder have to do with velocity and displacement that it would be that critical? If the servo is sized correctly, only the pressure is going to be affected by friction in a positive displacement pump. Over coming the static friction is going to be a function of the motor.

And yes, you can make your own piston pump out of a PVC pipe and a sealed plunger.

And wow, pushing water that fast through small check valves... sounds like cavitation to me. 4-inches in 300 ms? I don't think water likes to go through little valves that fast, judging from your drawing.

None of this sounds right to me, but that's just me... I'm sure someone will come along and correct me, they usually do.

Have you thought about cavitation?

Charlie

http://www.facsco.com

 

[ornerynorsk]

Are these pulses of water a closed loop system, or feeding a fresh charge of water each time it reciprocates?

It is better to have enough ideas for some of them to be wrong, than to be always right by having no ideas at all.

 

[johannestoger]

@ornerynorsk: It will be a fresh charge of water in every pulse, i.e. we'll fill the pump from a fresh water source on the inward stroke, and empty the pump into the experiment on the outward stroke.

@FACS: Thanks, great to hear some input from an expert. I should mention also that we need the setup to run for about 10000 cycles without servicing it.

You're right about the cavitation - the in/out valves won't be that small, sorry for the confusing drawing. I'll make sure to calculate the friction losses for the valves I choose.

The pulse duration will be 300 ms (milliseconds), and the peak velocity 0.6 m/s. However, I'm going for a slow acceleration and slow deceleration in a sinusoidal pulse. So the total displacement will be less than 100 mm, actually, since the piston is moving slower than 0.6 m/s most of the time.

I found some nice and cheap servos at

http://www.trossenrobotics.com,

they should be powerful and accurate enough for this application according to my calculations. Do you know of any other good servo sources?

With this in mind, (top speed 0.6 m/s, diameter 25 mm, stroke length around 100 mm, max pressure about 0.1-0.2 bar), do you think a PTFE pipe and a rubber O-ring to seal the plunger would work? The ability of the O-ring to prevent leaks and the wear due to the plunger movement are my biggest concerns.

Thanks again for taking your time to answer!

 

[FACS]

We have several sources for servos, but we make industrial systems all the time. I don't know where you are located so I wouldn't know the reps. You will also need a servo controls and someone to program it.

There are many types of "rubber" O Rings. For water, a typical Buna will do. For extended service life, I usually go ahead and design a double seal.

PTFE pipe may be too smooth. You don't want a perfectly smooth bore for dynamic O Ring sealing. The Buna will squeegee and end up ripping. In that case, you may want to look at different styles of seals and materials. We usually cross hatch the bore and use proper dynamic O Ring gland dimensions. For dynamic seals, a surface finish of 16-microinches is recommended, you can not go any lower than 5-microinches as the oring surface will be wiped dry. Always use the largest diameter (cross section) O Ring that your design will allow.

The cavitation would only be an issue on the return stroke when you are drawing water back into the piston, if you can slow down the refill that will help. You are basically moving around 2.6 gpm so those small orifices seemed to be a restriction. If they are not really that small then ok.

Why did you choose this method any way? Seems expensive, you need a good servo for that many cycles. Did you consider running water through a regular PD pump, through a valve and into a reservoir... then, when you want a pulse, open the valve for the required time. There would not be any recharge time, no piston design to play with, no expensive servo and no extensive servo programming. The pump will gladly run all day and the valve would last for a long time. A simple cheap PLC would pulse the valve, and could even monitor the pressure and flow rate. Giving you more data control.

Then again, I don't know what you are doing...







Charlie

http://www.facsco.com

 

[imcjoek]

The pulse duration will be 300 ms (milliseconds), and the peak velocity 0.6 m/s. However, I'm going for a slow acceleration and slow deceleration in a sinusoidal pulse.

Off the wall idea time:
Wouldn't driving your piston-pump with a crank (like a piston engine) produce a sinusoidal output? Then you just need a simpler DC gear motor with the right output speed.
1 revolution of your crank would be ~600ms (300ms output pulse, and 300ms intake), which is 100RPM.

Warning:
I didn't put much thought into this...