Using an O-Ring to seal on an NPT lead chamfer - Design Improvement Ideas

[Josh Blank]

Hello, long time listener first time caller.

My foundry uses a water immersion tank to leak test brass valves that we cast and machine. Temporary plugs are installed into the valve and 20psi air is ported into the valve through a special plug. If air bubbles through the cast/machined surfaces, the valve leaks and we throw it back in the furnace to try again.

The bane of my existence is that we use a male NPT plug with an O-Ring that seals on the lead thread chamfer. The NPT thread OD is a little overcut so the fit is a little loose (prevents damaging the threads, facilitates ease of install/removal). This works out pretty well for how sacrilegious it is. The only problem is that where the lead thread actually starts (where the first thread overlaps and becomes the second thread) we sometimes get some air that bubbles through, and the valve is incorrectly deemed a leaker.

We are looking for a way to improve on this design. We cannot seal on the face of the valve where this port is because we don't want to mask any leaking that could come through the face (we previously sealed on the face, customer complained about leakers in the field). We've tried using plastic NPT plugs to seal on the threads, and those worked great for maybe 15 cycles before they got too deformed/beat up. We would prefer not to use any pipe dope / tape since we leak test hundreds of these valves per day and the lost time would not make up for the occasional false leaker. Our rate of false leakers is maybe 1%, but I've never personally seen one so it may be lower than that.

The O-ring currently rests in a relief that's cut at about a 30 degree angle (parallel to female NPT thread). We've been kicking around the idea of changing that relief to be either a full square or a 60 deg chamfer to match the female NPT threads. Does anyone think that would actually help seal on the thread overlap, or just wishful thinking?

And then the big ask here is if anyone has any better idea for a quick-change, full seal NPT thread given the constraints of this process? I am pretty much at the point of collecting up all the potential false leakers, and doing the test again with a real NPT plug and pipe dope.

Please forgive my MS paint sketch, we only have one CAD license.

 

[lucky-guesser]

This is far from my expertise but I'll give it a go.

I suspect your middle drawing will be similar or maybe slightly worse than your current design. I would think that compressing your circular o-ring into a square profile would leave gaps in the corners of the profile, allowing for too much potential deformation and lack of seal.

Your right drawing in my mind would have a good change of sealing as you compress the o-ring against a relatively flat surface, and as it gets pushed down it will sit tight to the shoulder. Issue being this will likely stretch the ring over repeated uses. Makes your maintenance a little more difficult, but o-rings are relatively cheap.

That is all assuming you have things drawn at least somewhat proportional to each other. If your rings don't actually sit how/where you show then things could change.

 

[Josh Blank]

Lucky Guesser, your name gives me a lot of confidence. I'm going to try a few with the right-side diagram, but also I'm going to try the current and new design with softer O-rings, and various thicknesses on the o-rings. I think we're well off the beaten path here, so I appreciate your disclaimer but completely understand that this is all just a lucky guess.

 

[lucky-guesser]

this is all just a lucky guess.

Lol, feel free to take my comments with a large grain of salt.

 

[Prometheus21]

"This works out pretty well for how sacrilegious it is." - indeed sacrilegious! And on par with the rest of the industry

We had a similar problem in my company. Now for smaller series we just use thread tape, but for larger series we use square rings and x-shaped rings to seal at high pressures, similar configuration as to your inverted 60 degrees angle design (far right). Considering your low test pressure this should be sufficient. Remember, scratches and rough surfaces texture will always cause leakage even at low pressures, to combat some of this you can use a softer hardness rating (75 ShA instead of typical 90), as a softer sealing ring will easily seal minor irregularities in the mating surfaces.

Best of luck

 

[btrueblood]

We use a groove similar to your middle drawing, but smaller than a "regulation" groove per the Parker handbook, i.e. the ring is barely held in place, and the groove helps to stuff it into the last female thread chamfer. The result is a lot of deformation and wear on the oring, but orings are cheap. First instruction for a leak on the inlet/outlet is replace the oring. And agree that a 70 dur. ring seals better and lasts longer than harder durometers. Also, we have a clamp fixture for testing, and forgo threading the plugs in/out (and potentially damaging the threads) - the male part of the test plugs are a conical surface that matches the taper of the high points of the female threads.

 

[geesaman.d]

My limited experience with NPT threads is they don't engage a consistent distance. That's maddening. The first two designs will be sensitive to thread engagement to achieve proper o-ring squeeze. The third design should be insensitive to this and so it's my first instinct.

 

[Josh Blank]

Thank you all for your responses. We are first going to try swapping out the O-rings on the existing design with softer O-rings. If we were to go with the far right design concept, are there any concerns about not constraining the O-ring in an actual groove? Would the O-ring tend to expand radially since there is no ID to hold it in place? My machinist tells me that cutting the 60 degree chamfer would be fairly easy, but if we were to cut that into a groove we would require some special tooling to be built. That's not a deal breaker, but obviously money is an object.

 

[geesaman.d]

I assumed that with the right concept, there would be a smooth cylindrical surface on the mating part. Then the o-ring squeeze will be purely radial and the o-ring deformation will be purely axial (down)

 

[btrueblood]

You really need to sketch the mating surface (female threads) to see what is happening. The oring is going to bury itself into the first thread chamfer (assuming you have a start chamfer on the female NPT, and if not why?). You need something to hold onto the od of the oring and keep it from slipping outboard and leaking by, and also to stuff it into the chamfer of the female threads to seal. The middle idea does that best, imo.

 

[Tmoose]

An NPTF thread theoretically seals "better".
I suspect in practice sealer is still required.

An NPTF plug plastic made of a more robust plastic or elastomer might have a usefully long life.
Maybe a polyurethane ? Some flavors (cast polyurethane)used for O-rings are rated as having exceedingly high abrasion resistance.