Insane Hydraulics

Site theme image

Interface Leakage Caused By O-rings Working as Pumps

I said it before and I will say it again - technical literature available from the SUN Hydraulics website is top-notch! Their tips and guides are of the highest quality, and (if you haven't started already) I advise looking into their tech resources regularly because they are filled with nuggets of hydraulic knowledge.

Today I want to talk about SUN's guide on the mounting of sandwich bodies (the original can be downloaded here). It is a very informative piece that provides useful tips on the assembly of ISO03 sandwiches (which, by the way, in SUN's execution, use symmetrical design that allows you to flip them about, which is awesome in my opinion) - but one small paragraph of the document, devoted to interface oil leaks, caught my attention. Let me quote it here unchanged:

Quote "...Steps to Avoid Seal Plate Leaks:

1. Make sure the mounting surface is flat and without burrs.

2 . Do not polish the mounting surface. (Machining marks help to keep the O-rings from acting as pumps.)

3. Remove any dings on the corners of seal plates that could create gaps.

4. Studrods or cap screws must not bottom out in the threaded mounting holes.

5. It is very important to keep all of the o-ring sealing surfaces “dry”. The hydrostatic properties of an oil film under the o-rings, and at the seal plate interface, can create a fluid path leading from the pressurized port to the outside edge of the stack.

(Under this condition, as noted in step 1, an O-ring can actually act as a “pump”.) Warning: If you start with a wet seal interface in your stack, there’s a good chance it will always stay wet!..." End of quote

I hate oil leaks, which makes this part especially interesting for me! Two times in a row, SUN engineers mention that o-rings can "act as pumps" using hydrostatic properties of the oil film under the o-rings. I find this piece of information very interesting! Intriguing even. Because I've mounted countless face-seal-based systems, and honestly wasn't aware of the fact that mounting a wet o-ring on such an interface can create leakage issues.

I definitely changed the way I approach face seal mounts since I read the article, but I also decided that I would at least try to put this piece of knowledge through a real-life test when I get a chance - and a great opportunity to do this presented itself last week. So - I ran the test, and I will report on the results in a minute, but before I do so, since the guide, unfortunately, doesn't explain the wet seal leakage phenomenon, I would like to theorize a bit.

I imagine that a wet o-ring, even when compressed in its cavity, will allow for a thin film of oil to remain under it, at least for a certain amount of time, and if it is then subjected to a periodic pressure loading (which would be the case of any reciprocating industrial hydraulic system) - maybe a situation can emerge when the thin oil film would allow for a small amount of oil to be expelled to the outside of the o-ring periodically whenever the system pressure cycles. At least this is what comes to my mind when I read the phrase "o-ring acting as a pump".

So, to test this I would need to find a reciprocating load that "comes in an ISO3 (NG06) package", then mount it with "incredibly wet" seals, and then let it run for some time and see what happens. While I have a test bench in the shop and plenty of NG06 base plates and valves - the reciprocating part is more challenging, because I imagine that I would need to run such a test for at least a couple of hours to see if I catch any leakage, and I don't fancy working a manual DCV for that long or devising an electrical system just for the test.

Fortunately - a perfect opportunity appeared by itself when a client of ours asked for a mechanical reciprocating DCV for one of his projects, and given the flows that he was operating with, a simple NG06 flow-based reciprocating DCV would be the perfect choice for his application. Since we needed to run some tests on the vale anyway - it was the perfect opportunity to put the wet o-ring leakage thing through a test.

So here are the test tools - a normal ISO3 base plate and a mechanical reciprocating DCV:

My test tools
NG06 reciprocating valve
NG06 base plate

I mounted the valve with what can be described as an "atrociously lubricated" interface, connected a small hydraulic cylinder to the work ports, and then hooked it all up to our test bench. By the way - this is how this stuff is mounted in real industrial conditions when you are in a hurry. Usually, nobody cares if the base plate is wet or not.

atrociously lubricated NG06 interface
atrociously lubricated NG06 interface

Then I did all the tests I needed to do with the valve, and then let it run for two hours.

The video shows the valve covered with oil because I shot it right after the "wet" assembly, but I carefully wiped off all of it and dried the valve and the plate with a degreaser before timing the test.

Wiping and drying excessive oil
Wiping and drying excessive oil
Wiping and drying excessive oil

And, finally, two hours later...

Still dry two hours later
Still dry two hours later
Still dry two hours later
Still dry two hours later

Yep - the manifold was perfectly dry. Not a drop, damn it! I saw nothing that could even resemble a more or less identifiable leakage.

Then I thought that since I assembled the valve with the seals in the cavities, maybe the external part of the o-ring that was pressed into the cavity didn't get wet. So I removed the o-rings and made sure they are "properly wet" this time:

The o-rings are properly wet now...
The o-rings are properly wet now...
The o-rings are properly wet now...

Then I dried everything up once more, and let the "system" run for another hour and a half. And...

Still dry...
Still dry...
Still dry...
Still dry...

The valve and the base plate were, once again, dry! Note - the last two pics show a die-cast surface that is "touched" by oil, but that's not leakage. I believe it's the remains of the oil from under the valve when I assembled it, which had to eventually seep out because the whole contraption was at 50-plus degrees at that point. At least not a conclusive leakage.

Now, I absolutely believe that SUN's engineers presented a valid point. I just didn't manage to recreate the failure. And I wonder why.

Was the pressure insufficiently high? Is the leakage so small that it would only be revealed after days of operation instead of hours? Does this leakage only happen to seal plates, mentioned in the guide (as opposed to machined cavities)? Was my base plate surface place not polished enough?

Be it as it may - at this point I simply present the results of the test as they were. If anybody knows someone from SUN Hydraulics who can shed a light on this matter - please let them know about this post, maybe they will be willing to share what they know.

I am hoping next week I'll have an update on this subject, and a lot of us will learn something new.