> (1) They narrowed down the source of the failure using "acoustic triangulation", which, I think, is essentially using sound sensors (accelerometers?) located at various locations to pinpoint the location of the failure in 3D space.
An accelerometer is a poor device for doing acoustic measurement. At best, it can measure how the surface the accel is attached to responds to the acoustic environment (and that's only if the accel has a very high frequency response and is conditioned and sampled appropriately). For this kind of work they probably wanted to measure dynamic pressures within the tank, which they would most likely do with microphones. Other than that you have exactly the right idea.
Well depends on their bandwidth. An accelerometer is essentially indistinguishable microphone also capable of detecting sounds all the way "DC" sound (where 'sound' is applied force on the accelerometer itself rather than a pressure plate that reacts to air pressure) -- if it's sampling rate is fast enough and it is coupled to the structure (i.e. the coupling system also has enough bandwidth) it is an acoustic sensor.
Of course, a microphone was designed with high bandwidth in mind, while accelerometers probably value more precision around the DC input.
Not all accelerometers have usable response at DC. The accels I have on my platform for dynamics measurement (high structural vibration and flutter) are meant to be AC coupled, for example. At rest they may give some garbage reading, but when they're vibrating, they accurately measure how the structure is responding.
Your point remains though, if the accel has appropriate frequency response and is properly conditioned and sampled at a high enough rate, acoustic measurement is possible. What I was trying to say (maybe poorly) is that an accelerometer would not be my first choice for doing such work. But in the aftermath of an event like this, the data may be (and was, it sounds like) usable for such purposes.
An accelerometer is a poor device for doing acoustic measurement. At best, it can measure how the surface the accel is attached to responds to the acoustic environment (and that's only if the accel has a very high frequency response and is conditioned and sampled appropriately). For this kind of work they probably wanted to measure dynamic pressures within the tank, which they would most likely do with microphones. Other than that you have exactly the right idea.