Around 10:30 Monday night, May 3, a train on Metro Line 12, Mexico City’s newest, was traveling east on an elevated portion of the track from Tezonco station to Olivos station. Line 12 was fraught with problems during and after construction in 2014 and 2015. Several sections of the line had been shut down for repairs in 2014, and some citizens complained that the line was further damaged by the 2017 Puebla earthquake.

Just as the last two cars of the train were passing over one of the Y-shaped concrete support structures that hold up the girders supporting the tracks, the support structure failed. This caused the girders to collapse to the freeway median underneath the track and dropped the train cars some five meters to the pavement. A total of 24 people died either in the wreck itself or shortly afterwards and 79 people were hospitalized. It was the worst Metro accident in over 40 years; the Mexican government declared three days of mourning for victims of the disaster.

Mexico City Mayor Claudia Sheinbaum has invited a Norwegian accident-investigation agency to look into the causes of the tragedy. Definitive answers will have to await the agency’s report, which may take months or years. But there are some things we can say already about what happened.

Every engineered artifact has both a physical aspect and a human-relations aspect. The physical aspect is simply the thing itself: in this case, the concrete supports and girders that failed. The human-relations aspect (an inadequate phrase, but I can’t think of a better one at the moment) is the history of the human organizations, relationships, and interactions involved in the artifact’s design and construction. The human part of the equation is a “structure” that is equally as important as the physical structure it produces. Both the physical objects and the human-relations history have to be investigated for a complete picture to emerge about how things went wrong and who might be responsible.

As every civil engineer knows, concrete is strong in compression but weak under tension. You can squeeze it and it resists well, but a pure concrete cylinder with no reinforcing bars (“rebar”) in it will fail pretty fast if you pull on it. That is why support structures such as the flat-Y ones that held up Line 12 have to have extensive rebar networks inside them to handle the tensile stress that shows up on the top of the Y, as the pressure of the girders try to stretch the arms outward and downward.

I am not a structural engineer and I have exhausted all my knowledge about reinforced concrete in the preceding paragraph. But people paid to know these things know how to design such structures so that they can withstand both static loads and also the dynamic loads of things like moving trains and earthquakes.

Mexico City is one of the most earthquake-prone large metropolitan areas in the world, and one would expect their engineered structures to show evidence of this fact. Without additional information, I can’t say what special precautions have been taken to ensure that the Line 12’s support members could withstand earthquakes. The 2017 Puebla quake had its epicenter near Mexico City and a magnitude of 7.1; it resulted in over 360 fatalities and the collapse of some 40 buildings. Although the newly-built Line 12 withstood the quake without collapsing, it is an open question whether the quake might have caused hidden damage to some parts of it.

Admittedly, finding internal flaws in concrete structures is a hard thing to do. Although non-invasive technology such as ultrasonic testing and X-rays can be used, they tend to be either expensive or inconclusive or both. The only structural test that many civil engineers will accept as conclusive is to stress something until it breaks. That’s fine with small test samples, but it has obvious drawbacks for testing completed structures.

Recent research, including some at Texas State University where I teach, has been focused on building “smart structures” that incorporate electronic sensors which can alert engineers to incipient flaws before they get serious enough to threaten the structure’s integrity. If something like this had been installed in the support members on Line 12, it’s possible that authorities would have known about the defective support well before it failed and could have taken steps to repair the defect.

But in the absence of smart sensors, the only way to prevent such tragedies as the one that happened on Line 12 last week is to build supports to be strong enough not to fail. And here is where the human-relations issues come to the fore.

A Reuters article on the accident points out that the portion of Line 12 that collapsed was built by a consortium of a Mexican business empire controlled by the family of famed businessman Carlos Slim and the Mexican division of a French firm. Large engineering firms can be as good or better than small local firms, and bigness by itself is not a vice. But governments and businesses in Mexico have been known to host corruption problems. And any time a public work such as a Metro line spectacularly fails, it is time to scrutinize the history of how the work was contracted, how it was inspected, and whether any shortcuts or graft were involved.

As the contracting firm itself pointed out, it is premature to assess any blame or jump to any final conclusions regarding the cause of this multiple-fatality accident. And Mayor Sheinbaum may be doing the right thing by calling in well-respected experts from another country to perform the investigation, if local or federal inspection services are not up to the job or could not be expected to deliver an unbiased report.

So we will simply have to wait for the investigators to draw their conclusions. In the meantime, Line 12 remains idle and thousands of Mexico City commuters have to find another way to get to work. But at least they can be reasonably sure they’ll get there.

This article has been republished with permission from Engineering Ethics Blog.

Karl D. Stephan received the B. S. in Engineering from the California Institute of Technology in 1976. Following a year of graduate study at Cornell, he received the Master of Engineering degree in 1977...