dynamic rotating surface seal help

[jrkeme]

OK I'll try to keep this brief.
I have an application that is meant to diminish (not seal) leakage from a 50-100psi vessel around a 5 inch rotating disc out an orifice. So within the pressure vessel is a 1 inch rotating shaft with a 5 inch disc mounted to it. The disc at the moment is right up against the end cap housing. We tried to get a very small clearance but it still leaks too much. So this rotating disc that is right next to the end cap of the pressure vessel is rotating with 2 holes in it. As the holes rotate they uncover an orifice in the end cap (that is what they are supposed to do) However since we do not have sealing around the outside of the disc there ALWAYs a permanent leackage path to the orifice in the end cap. What we want is for higher pressure air to only go through the orifice when one of the two holes in the disc align. Not that PLUS the leakage around the disc.

SO! What I am looking for is not shaft sealing. What I would like to do is put an "o-ring" or equivalent around the perimeter of the disc so that when the disc is pushed up against the end cap (recall we tried to do 0.010") there is some type of sealing mechanism. So think of the disc as a hamburger with a 5 inch onion ring around the perimeter that is "compressed against the end cap. I am very ok with this NOT being a actual full seal. It can be a clearance seal that as the "material" wears due to "break in" the remaining gap is very tiny. This is my cave man view of what I think will work. ESPECIALLY Since what I would like does not seem to exist. Keep in mind I cannot easily put a "lip seal" around the circumference of the 5 inch disc against the radial part of the housing. That gap is greater than 1 inch.

Does something like this exist? To conclude I am expecting some leakage. Is there a consumable part/ring that I could install into the rotating disc that when pushed up against the end cap I get an effective seal to leakage around the outside of the disc. So really my onion ring is fully in contact with the end cap at all times. I was thinking of something made of PTFE or carbon?

Thanks everyone
Jason

 

[jrkeme]

MintJulep
I expected that from you.
I would think a rational person would attempt an alternative solution before scraping the entire project. Perhaps you are made of money or work for a place that prints it, but I do not. As such I will attempt to look into feasible solutions before simply tossing everything in the trash

 

[Orange_kun]

I think you need to be more explicit about what the requirements are.

First it was a 2 hole plate. Now it's 3. First is was 1000 rpm, now that's not the test speed. All this for a test rig? How long is the test actually going to take? Can it take longer and run at lower speed? You mentioned a possible solution is just raising the input pressure. Why wouldn't you do just that and forget about any redesign?

Finally, the PPT slide shows the mating plate as "PTFE" as well. Why not machine a huge, shallow, spot face on it for the rotating steel to fit into. Grit blast the steel and let it wear itself in to a close fit on the large face and a shallow radial face. Just a thought.

I don't love any of the seal designs over a dedicated valve. What happens if the seal wears further during test and the leak rate rises? How does that affect test results? Is this all occurring with 20C air or are things getting hot? MintJulep's comment about sunk cost is also extremely valid. How much is your time worth? And a custom part? And the time to characterize the new solution? Sometimes these hack-n-slash solutions work great. Other times they are a continuous source of frustration for technicians. Might be worth a revisit.

 

[jrkeme]

Thank you Orange_kun. and I agree with your comments. I am well aware of sunken costs. However I don't have an army of machinists at my disposal nor the cash to redesign a new system let alone the time. The upper target was 200 orifice opening events per second. Very possibly a stretch goal. Some gas dynamic modeling showed that it could be possible. However a more likely target is 100 opens a second. That is shy the disc I currently am using has 3 holes so at 33Hz I would get 100 openings a second. Again these are not strict requirements. What I want to do is cycle pressures as fast as I can. The device is a test rig for a specific component. That under normal operating conditions may run for several years. Thus the test apparatus is meant to accurate the cycling not to mention the rig is about 1/50th of the size of the full machine. The PPT above was just a sketch of some possible solutions where right now there is no sealing just close machined tolerances with what was modeled as acceptable leakage rates. However in actuality the leakage is quite a bit higher. I am simply trying to knock those leakage rates down some

 

[3DDave]

I have seen huge amounts of cash wasted on a bad concept. The worst part is it delays starting the search for a concept that will work, but it starts with a suitable problem statement.

In this case I see the need is for a continuous bleed of air with pulses of some duration and trying to make a large volumetric flow.

Each moving hole will produce a zero to maximum to zero contribution of 3 diameters of rotation, about 30 degrees by estimate; for 3 holes that is a total of 90 degrees or 1/4th of the rotation time. At 100Hz, that will be 2.5 mSeconds total duration for the three holes or 0.8 mSeconds per hole.

That seems like an easy amount to just be averaged out like with Pulse Width Modulation; not delivering pressure spikes.

 

[MintJulep]

Interesting. You refute my pointing out that you are succumbing to the sunk cost fallacy by claiming......that your sunk costs are too great to stop now.

Over my career I've seen lots of people do exactly what you're doing now. Falling in love with their first idea and sticking with it, even in the face of their own objective evidence that their idea doesn't work. I've never seen this path end well - but perhaps you'll be the first.

What you are building is similar to a "rotary timing valve". I've used them. I've designed modifications to them, so I've had to figure out in detail how they work. To get close to zero flow at cut-off the faces need to touch, so they need to be super flat, super smooth, have low friction and good wear properties. You seem willing to accept non-zero flow at cut-off, so your disks don't need to touch. That does make things easier.

Except, as Orange_kun points out, your velocity at the edge is non-trivial. So you still need flat, smooth and low friction. Then, you need to find a seal material that can deal with the heat it will generate and last as long as you need. You'll have lots of flow, so that might help to keep things cool.

The force across a 5-inch diameter disk with a 100 psi pressure drop across it is not huge, but it's not negligible either. Because of the holes and rotation, the force will be cyclic. So your next challenge will be with disk stiffness, bearing stiffness and shaft stiffness.

If you need the cut-off flow to be "small", then the disk faces need to be close. The system stiffness needs to prevent the disks from crashing.

Since you seem willing to accept a cyclic pressure of not-zero to peak, then as 3DDave suggests, one steady state flow in parallel with a modulating flow (easily achieved with a solenoid valve) will get you there.

If you want closer to zero to peak, then just connect your thing under test directly to the output of a single-stage reciprocating compressor.

You haven't shared any real requirements yet, and what you do share keeps changing. That makes it hard for us to provide any more meaningful commentary than we already have. You will get better advice by telling us what you need to do, not how you think it should be done.

 

[jrkeme]

Thank you MintJulep for your constructive suggestions.
Yes I know the FPM on the edge of the disc is high but not overwhelmingly so. Somewhere around 2500fpm.
The reason I have not provided exact specifications as most are "the fastest you can get type" Gas dynamic modeling seemed to indicate (with plenty of assumptions on leakage rate) that it was doable around 100x per second (opening events not shaft speed)
Yes zero to peak is not necessary. that much is certain and something not obtainable with the current rig. Which is why I'm not completely abandoning this concept either. The solution I am looking for and the reason I made the original post is "Is there as COTS seal that could be used to upgrade the rig (already built) to improve the leakage issue to a more manageable level?" Modifying the discs to accommodate a COTS seal (of some type, when I originally posted I was struggling with the right terminology to even look into) was relatively inexpensive and from a time perspective worth looking into for a team of 2 for a month. (not full time) Should that not meet with an acceptable solution then a redesign would be needed.