So, what you're saying is that a product designed for a specific purpose is not to be trusted for its specified purpose. Therefore it is worthless and the engineer who designed it is also worthless.
Do you know what machinists think of engineers?
I have a 15 psi system that I need to seal. I had a 95% success rate with the product I used that was supposed to be good for 100+ psi and 2100 degrees. I figured it was overkill, but I was wrong.
Fadal, didn't mean to imply that the design is cr*p, or that you engineer is an idiot or anything. And I sense from your reply that this may be a situation in which you have 5000 of these things that are already made and ready to go out the door as soon as you solve the sealing problem. I understand that my comment may have induced stress. Sorry, I was trying to be helpful.
Whatever anyone's sentiments, straight threads without gaskets on the face are not in general made to seal stuff. That you get 5% failure with a very high-performance thread sealer (2100°F! Wow - what sealant is this?) testifies to this fact. My note was to suggest that a minor design change will make your life easier. The shuttle had a problem with seals. No one would say that the shuttle is a pile of cra**y engineering. They redesigned the seals, and shuttle performance improved.
My suggestions were to help you. A five percent failure rate for any part of any manufactured product today is so far from acceptable it's not funny. If this product is "standard", for standard autos, and you sell 10,000, field servicing 500 units is gonna be unpleasant. And it it's high-end, and you sell 100, having 5 customers out there dissing your product in a small community is bad for business. And even if you get something that works in your shop, applying even the best threadlocker to an improper (straight thread) design could leave you up at night, worrying about a failure in use.
Back to your question. You want a way to ensure few leaks in an automotive environment. If this is for super high temperature and the user will need to disassemble and reassemble this occasionally, I'd figure out a way to use a crush gasket. That's probably significant design and engineering, but it's proven in automotive use. For more normal temperatures, the o-ring boss design seems to me a reliable, simpler, and cheaper way to go. It may mean an extra counterbore and grooving operation, or you might get away just by making your thread relief groove a bit deeper. Or you could silver-solder or braze a Swagelok fitting in the hole. Or you could change the threads to tapered pipe thread. Gbent's excellent description points out that this will give you a much more reliable mechanical seal. A taper with sealant should be a joint you never have to worry about. I personally don't like taper threads for stuff that is frequently reconfigured or disassembled, though.
I hope that this is taken in the spirit it was written. If one reads the responses on this forum you will see that if you ask "Should I use product A, or product B", your answers will range from "Use product C" to "That's not the way I would make it. I'd use a Qwernsdorfer Bifurcation Indexable Step-taper machine, with the Nermbly attachment. It's perfect for this operation". Or, someone will chime in with a story about their Uncle Al, or something. Usually, these comments are made to be helpful, I think.
Good luck,
J