I think there's still some confusion as this is incorrect:
Old method: Kettle wall/O-ring/electrical box/sealing surface
Look at this picture of my design:
The electrical box cover that sits against the o-ring is so thin that the seal in my setup is indeed the same as yours which is:
Kettle wall/O-ring/sealing surface/electrical box
The o-ring and washer are chosen such that the o-ring completely compresses against the flat side wall of the heating element as well as the kettle wall and the electrical box. This works because the cover plate is extremely thin (0.92mm).
The only reason to remove the element is if it's damaged and needs to be replaced. You mention that your method of welding it to a plate makes this easier but I'm not sure how as if you need to replace your element, it's not only $20 for a new element but you need to re-fabricate a new mounting plate and weld the element to the plate. For most people, that's more work and expense.
As others have said, the silicone on the inside isn't needed. Even if you do use it and need to replace the element, cut it off with an exacto and re-apply more later if you like. One tube will last you for many many elements.
The very thin element box cover on the inside bends somewhat (as does the thin washer) when you tighten things down which allows a good seal to be created against the kettle wall. If the plate used were rigid, this would not occur. On your setup when you tighten down the top and bottom of the o-ring will compress more than the left/right edges since the plate is thick/rigid. You can your o-ring here:
Since there's no washer around the o-ring on your design you have to be careful to not overtighten and cause the o-ring to bulge (causing leaks) nor can or should you bump or move the element box. That to me means it's more prone to leaks. This is common concern and complaint that people have always had with traditional weldless valves that use a silicone o-ring for structural support. They ask that you don't overtighten. But then they leak when you twist and turn them sometimes but you compress/decompressing the o-ring by varying amounts in different spots, causing leaks.
When you mocked up my design, are you sure that you didn't have the heating element rim inside the thick electrical box? It needs to be only against the thin cover, which then only reduces the element thread depth by 0.92mm (0.0362").
In other words, is the same box you used when you mocked it up?:
If yes, that won't work. The hole needs to be much larger. It has to be like this so that the element sits flush against the really thin cover. This is mine:
0.92mm is a very tiny amount, likely less than the thickness of the washer you used in the inside of the kettle. Especially when you consider the air space you have between it and the kettle wall on your setup:
If you could not get my design to work by reducing the element thread length by only 0.92mm then you did something differently than I did. Or maybe the element thread length is shorter on yours? Or something else is different?
Either way, if you got it to work then that's all that matters. We'll just have to agree to disagree which method is more leak-proof.
Kal