I don't know much about the specifics of gas distribution, but I came across a comment on an entirely different discussion recently which I think is apropos:

> you should be very suspicious of any conclusion that requires you to assume that all the world’s experts have missed something extremely basic.

It seems to me most likely that the answer is simply that releasing natural gas into the air is an even more dangerous failure mode than overpressure - natgas is not air in a compressor. But regardless I'd bet you dollars to doughnuts that, for one reason or another, blow-off valves are a bad idea in the context.

I’m not sure I agree with the statement—humans overlook basic details all the time. Most computer bugs are “extremely basic”. Multiply times possible failure points and it’s easy to see how basic flaws can easily cause systemic failure. Experts DO miss extremely basic things daily. You have to actively build proccesses to avoid this. Expertise is not enough!

> you should be very suspicious of any conclusion that requires you to assume that all the world’s experts have missed something extremely basic.

This is hugely mistaken. If you're an expert in compliance in one field, often your insights are valid in another field.

For example, anybody who has certified a data center knows that Fukushima was both in the wrong location and had wonky power distribution.

Anybody in aviation knows the Tyndall F-22s were in the wrong location after Kermit Weeks' collection was destroyed in Florida in 1990.

Expertise and common sense are always in short supply. It's 2018 - we've collected enough hindsight for the next 100 years.

You could flare it, but you would need some combination of space or a high enough tower.

Use a pilot light to ensure it can always function after the release valve/disc blows.

I see your logic, but the fact that overpressure demonstrably caused a catastrophic failure sets the bar pretty high for blow-off having worse consequences.

Compressed air can be released back into the atmosphere from which it came without much danger, but gas is not so harmless. Relief valves exist to prevent pipes from exploding due to the pressure, and during this incident they were probably wide open, but they just continued to release gas which eventually exploded anyway.

> it seems to be a pretty stupid design that doesn't have relief valves.

They have, the problem here was that the relief valves vented way too much gas - they're in the houses of people... and one spark with the right amount of oxygen and stuff goes kabooomm. Gas fires are no joke, gas explosions even less.

There are pressure relief valves, but they (a) can't handle that much overpressure and (b) many of them are located inside the houses (at the gas meter, which on many houses around here is still indoors).

This is why all the new gas meters in SF have their relief vents piped outdoors - much lower risk of explosion. If you see those little squareish-looking pipe ends with a fine screen mesh on the end that’s what they are.

Why not install the new gas meter outside? Or do you mean replacement gas meters?

The usualy idea of the valves is for them to open before the overpressure exceeds their design capacity.

A blow-out disk isn't going to fail to operate. My pressure-cooker has one. It's just a piece of material that will catastrophically open the vessel to the atmosphere before massive overpressurization.

Well...that's a very unfortunate example. Pressure cookers' blowout disks can get clogged by whatever you're cooking (beans specifically) and the whole lid blows out (embedding itself in the ceiling and painting the kitchen). Scary stuff.

I have a Presto pressure cooker. The blowout disk is a piece of rubber/plastic built into the cap. I have a hard time imagining it getting contaminated in a way that surpasses the strength of the stainless steel lips on the lid/pot.

Its design hasn't changed since 1977.

It's not a tube that a bean could get into. Maybe you're thinking of the vent pipe? When that gets clogged, the overpressure plug is what blows away to release all the pressure.

Well...I have seen the whole lid, firmly embedded in the ceiling. IDK what the brand was, but apparently the failsafe (a rubbery thing, off-center of the lid) failed to fail-safe in that case.

Looking into RAPEX, I see multiple recalls for this specific issue (pressure buildup leading to uncontrolled blowout); all for brand names that are unfamiliar to me.

Of the 12 recalls I found on that database, none specifically mentioned failure of a blowout disc/plug1.

1 depended on the rubber gasket itself as the blowout valve. Several could be opened while under pressure, and some were deemed structurally deficient, which I take to mean that the vessel wasn't (consistently) built to withstand the design pressure (and a margin of safety).

[1] 1 did reach 290kpa (42psi), nearly triple the typical pressure cooker. Not sure what happened on that one.