Now I'm thinking that a weak part of the tank gave way at one of the lesser welds on this non-pressure vessel, quite early and in the top part of the tank well above the liquid level. When you stand on top of a big tank, a lot of them do flex quite a bit. Physical pressure must have increased somehow in a way that could not be relieved without rupture.
IIRC there was some report of a "gas leak" to begin with and that could very well have been it.
I can only imagine what it's like to be one of the shift operators down there on the pavement around the tanks and pipes on a daily basis, in their flame-retardant coveralls and hardhats. Based on my vivid experience working with similar operators as I took official samples from tanks usually located in much bigger tank farms.
MMA smells so sharp that any leakage from a valve or pipe fitting gets attention much more so than other flammable liquids, for instance things like paint removers, alcohol or acetone which smell very faint by comparison. I expect that the workers had gotten accustomed to a fairly clean air environment without serious toxicity levels, and would estimate that a leak of about 1 drop per minute from a flange would be easily identified and fixed. Not only to keep the neighbors from complaining, but to keep the workers from gagging when they walk closer to that location.
If the first sign of trouble was physical tank bulging and/or suddenly stinking up the whole place with lots of MMA vapor (like from an upper tank rupture) when there is no sign of leakage on the ground, that pretty much qualifies as a "gas leak". Once the smell was no longer prominent, I would say that polymerization was well underway and there was not as much free monomer available to build up more pressure or even evaporate very much after that. If there's warm liquid MMA open to the atmosphere, you will smell it downwind.
It might be a stretch to say that "nature takes its course" when synthetic monomers spontaneously polymerize like this, but that's what some of them do best that sets them apart.
From what I understand, in the uninhibited pure monomer it doesn't take that many PPM of monomer molecules to become "activated" in some way before the chain reaction of them with remaining molecules eventually accelerates to completion. Which may be why such a low concentration of inhibitor works so well, but inhibitor must be well-dispersed within the liquid and there must be enough circulation to avoid local depletion of the inhibitor due to stagnant contact with as yet unidentified unexpected initiators. Not only should it be easy to get a random 1 liter sample from a tank (of such non-viscous material) at any time, the single sample should ideally be highly representative of the entire tank contents. Due to the way the liquid is handled in bulk going in & out of the tank, as well as any additional circulation to insure good mixing of incoming cargo, or after boosting inhibitor amounts when only a few carefully measured pounds or kilos of the concentrated inhibitor is dumped into the bulk monomer to keep it from going below spec.
The inhibitors have been selected to be very strong and effective at what they do so it doesn't require a high concentration, and the downstream processors can more easily overcome the inhibitor effect and produce plastics using different and milder polymerization initiator techniques when there is not an excess of inhibitor.
Once it gets polymerizing past that minuscule little inhibitor, from that point it's like there is no inhibitor at all. So there is some possibility that this is a fairly worst-case runaway polymerization event where the full amount of caution was rightfully warranted. Even though it would have to be considered fairly benign compared to a whole tank of the liquid monomer dispersing through the neighborhoods, or going up in a fireball, which is always a risk anyway :\
Under storage, it really is tremendous force that the inhibitor is holding back, considering how badly outnumbered it is by the 99.9% of everything else in the tank that wants to spontaneously react with itself otherwise. When the inhibitor is only about 20 PPM, which in percent is 0.0020%, in these cases percent by weight. IOW 20 lbs of MEHQ crystals dissolved in every million lbs of MMA liquid, or in metric 20 kilos of inhibitor for every million kilos of cargo.
Now usually the lowest inhibitor concentration will be at the chemical plant that produces the monomers, and it may go into bulk storage the same way when fresh material is moving through the system adequately. From there, vessels, rail cars and trailers often have different amounts of additional inhibitor added in order to boost levels according to purchase agreements of the different industrial consumers.
Often by contractors such as companies I have worked for who store the inhibitor concentrate as agents for the client, and can sample industrial tanks & test in the lab to routinely determine inhibitor concentration. Then carefully measure according to requirements and be the third party that certifies the new concentration after our field operator physically adds the additional dose to the tank.
For something like vessels and rail cars you just can't expect their cargo samples to accurately reflect a recently boosted inhibitor value, the small amount of inhibitor concentrate won't be adequately mixed with the rest of the bulk monomer before the ship leaves the dock or the train leaves the terminal. So it has to be certified by careful physical addition while being supported by good lab data to begin with. You expect lab readings to match the certificate after the ship gets overseas or the rail car gets across the country.
Now measuring low PPM is not always easy but you get better after a while.
Much better for me because I got to start doing it my own way as soon as I could, after first establishing as many of the predecessor techniques as possible for reference, but people still wanted more reliable results and so did I since the beginning. Ever since then, doing the continuous improvement thing for more decades than you normally get.
Things I will work with, repeatedly ;)