7 ms·

Sinking of the US Cargo Vessel El Faro: Illustrated digest [pdf]

88 points by curtis 8y ago

markonen 8y ago
The recent Vanity Fair story about this tragedy is absolutely riveting, even though you know exactly what is going to happen: https://www.vanityfair.com/news/2018/04/inside-el-faro-the-worst-us-maritime-disaster-in-decades https://www.vanityfair.com/news/2018/04/inside-el-faro-the-w...
dmckeon 8y ago
All else aside, the UI/UX issues here are heart-wrenching. The captain preferred a graphic representation, even though it was 6 to 12 hours out of date, over a more current text-based report/forecast. On two occasions, warnings or new data arrived just a few minutes after a report/forecast was viewed. Crew members watching the Weather Channel had more current/accurate information than the captain. I can easily see how a graphic might be preferred over text, regardless of data currency, and especially during the 95%+ times of smooth sailing, but to see how this situation turned out is troubling.
- vpribish 8y ago
  I'd like to see the actual BVS UI. Does it indicate how old the data is or just the timestamp the report was generated? This is a detail that bites lots of reports in business.
  - marvin 8y ago
    Maybe the people who develop the software don't talk to the end users, or see how they use the software. That would be a classic example of development process gone wrong. Are they, or people sufficiently near to them on the organizational chart, aware that life-or-death decisions are made on the basis of their software? (I'd certainly hope so, plotting weather data definitely qualifies). I'm only conjecturing here; there's no reason to conclude based on the available information that there is a problem here. But it would be a pretty typical situation in an organization that believes software is something you call up a subcontractor to bolt on -- most places are still run like that, although the culture is slowly changing.
    - ethbro 8y ago
      Imho, that's one of the greatest value adds we (software engineers) can bring to the table. The "offloading scope / spec requirements to a PM" trend is a slow motion train wreck. If you, or at least someone else on your team, can't clearly state who's going to be using the code you write, what problem they're using it to solve, and how they're using your code to solve their problem... you need to hire an engineer who can empathize with customers.
    - AnonTemp123 8y ago
      For obvious reasons can't comment too much, but it does, they did, and makes quite clear that it's not for navigational purposes.
      - marvin 8y ago
        Interesting, thanks for the tidbit of inside information :) You'll have to pardon me for asking a stupid question, though -- what other purpose than navigation would a maritime weather report/forecast reasonably have? This seems to me at first look as a protection against liability, rather than an actual clarification of the intended usage of the product. I guess I could imagine a use case such as operations planning, e.g. "will we likely be able to perform work on deck tomorrow?", where the stakes are lower, but I imagine the potential for misunderstanding and incorrect usage of such a service would be dangerously big.
        AnonTemp123 8y ago
        You're welcome. It's for optimization of future routes -- how do I get from point A to point B while keeping some weather parameters in acceptable range, burning less fuel, and getting there on schedule. And yes, trying not to drive right into a storm. Not quite for trying to maneuver out after you have already driven into one. Of course you can also look at it and figure out of cleaning the deck is a good idea or not. I probably shouldn't comment further, but NTSB's full report is at https://www.ntsb.gov/investigations/AccidentReports/Reports/MAR1701.pdf https://www.ntsb.gov/investigations/AccidentReports/Reports/... and they did not seem to find it too confusing.
  - pcunite 8y ago
    It is now StormGeo and looks really nice. No wonder the captain he preferred it. https://vimeo.com/234734651 https://vimeo.com/234734651
- hinkley 8y ago
  One of the things I’ve learned from doing UI work is that some people absolutely do not picture things in their head. My boss, a generally smart guy, would insist we could squeeze one more thing in somewhere and even though everybody said it wouldn’t fit. So we had to build the stupid thing to show him it wouldn’t work. But until then he needed this thing to happen and neither physics nor his most capable people were going to convince him otherwise. Now, clearly this is a control freak we are dealing with, but also someone in authority who doesn’t trust images or shapes in his head. I had never encountered such a clear case of this happening and it opened my eyes to it happening elsewhere. Some people have low spatial intelligence. You would think becoming a ship captain would cull out a lot of people with that problem but maybe if you’re good enough at other things it doesn’t slow you down that much? Maybe he had to see things to understand them.
- Fins 8y ago
  From that mentioned Vanity Fair article I sort of got the impression that the captain expected the storm to be similar to what he had experienced when sailing Alaska, not what a proper tropical storm is.
curtis 8y ago
Something that I'd missed in previous articles about the El Faro sinking: > Then, at 0554 , the captain ordered a turn to port to get the wind on the ship’s starboard side, generating a port list. A few minutes later, El Faro lost propulsion --- > Log books indicated no more than 26 inches of oil in the sump in the months before the accident ( a ). With this oil level, the bellmouth would not take in air (lose suction) with an 18-degree list to starboard ( b ), but would lose suction with an equivalent list to port ( c ). They had other serious problems, but loosing propulsion in heavy seas is very serious indeed.
- oh_sigh 8y ago
  They would have all lived though if they had modern life boats that were enclosed and could be launched from a list
  - tgsovlerkhgsel 8y ago
    Are these boats able to reliably withstand such a heavy storm (assuming they don't get smashed against the ship or some rocks/land of course)?
    - lb1lf 8y ago
      Yes. The biggest issue with finding yourself inside one of these in a proper storm is that you’ll get thrown about violently (the lifeboat; you’ll be securely strapped into your seat using a five-point harness, or at least you were in the ones I’ve had evac drills in) for long enough to be as violently ill. I’ve been told by instructors - no guarantee it’s true, that, but imho sounds plausible - that the #1 cause of death in lifeboats or rafts after successfully abandoning ship is dehydration; you sweat profusely, puke repeatedly until your body can’t take it anymore.
- sjburt 8y ago
  It's interesting that the low oil pressure cutoff (designed to prevent engine damage from running without lubrication) is amongst the things that ultimately doomed the vessel.
mcguire 8y ago
Anyone know if the BVS data has a timestamp? An excellent and very readable document, by the way.
- lb1lf 8y ago
  -It does; however, the timestamp only shows when the report was compiled; it does not show when the data the report is based on were collected.
  - princekolt 8y ago
    As someone who builds software that generates exportable reports, I can say this is one of the cardinal sins of data visualization. Another one is relative timestamps akin to "just now".
    - Treblemaker 8y ago
      I know, right? https://pasteboard.co/Hn35e51.png https://pasteboard.co/Hn35e51.png
ocdtrekkie 8y ago
Definitely impressed by the design work that went into this report. I'd be super curious to hear from people who design these sorts of communications. Usually we think of a well designed pamphlet for marketing purposes, but clear informative presentation here well exceeds what I've traditionally expected from a government report.
- njharman 8y ago
  It is marketing. 1) they want all to "buy into" their proposed changes. 2) they want all to know how important their work is and why they should remain in the budget.
  - tlb 8y ago
    That level of cynicism may apply to some government departments, but agencies like the NTSB are much more grounded. When there's a bad accident on a ship that was more-or-less following procedures, owners and captains of ships really want to know what went wrong and why, so they can avoid it themselves. This is the sort of document that captains will read and discuss with their crew, not all of whom are highly literate in English, so it needs to be accessible and clear. It seems to me that the NTSB did good work here.
mathgenius 8y ago
I'm curious how much this is about "many small and unlikely failures leading to disaster", versus "they sailed into a big storm and were doomed". The usual narrative with complex engineering disasters is the former. I'm wondering, for example, if everything else was going well with the ship, could they have survived such a close encounter with such a big storm? Or, if the engine was kept online, would that have changed anything? It seems to me that this situation went far beyond survivability, but it's hard to tell from the NTSB report.
- BrentOzar 8y ago
  The fact that the NTSB made 63 different recommendations tells you that it was a complex situation. Their goal is to increase the odds of survival - and 63 recommendations means there were a lot of things that could have turned this situation around just a little. Completely survivable? Maybe not, we’ll never know - but 63 different recommendations means this was a really bad combination of things the crew couldn’t have overcome on their own. Contrast this with a typical plane crash investigation which produces, at worst, a dozen or two recommendations - that tells you a lot. I would have thought the maritime industry would have been better prepared. So sad what the crew was up against.
- rdiddly 8y ago
  It seems like both... don't sail into a hurricane and there's no problem, right? But even their sailing into it was an interaction of many factors -- the captain's insistence, predicated on preferring the graphically vivid but outdated weather maps, and on disregarding the opinions of his officers, and on each of them independently failing to be sufficiently assertive (or even convinced of the danger themselves) to change his mind. There's also the broken anemometer, but I'm tempted to discount that, because it was obvious the wind was from port, indicating (in any counter-clockwise-rotating storm, i.e. any cyclone in the northern hemisphere) you are headed right into it. Winds should have been from astern if you were trying to cross in front of it. When you're convinced of something, sometimes you can stare right at contrary evidence and disregard it. But who knows, maybe having an accurate wind reading would've jogged some brain cells. If the engine was kept running as you say, that might have helped them, but only if they had used it to turn and get the hell out of there. If the captain insists on using it to continue driving straight into a hurricane, it just makes things worse. And then of course there's all the water they took on, limiting their getaway speed so to speak. In a certain way it seems like the point of no return was the rupture of the fire water inlet. No telling what the ship could've withstood if they had secured the load better and secured the hatches better. Unfortunately all you get with this kind of hindsight is all the things that went wrong, not necessarily what would've had to go right.
wildrhythms 8y ago
This was an incredibly fascinating and shockingly well designed doc. Excellent work from the NTSB putting this together.

88 points by curtis 8y ago

The recent Vanity Fair story about this tragedy is absolutely riveting, even though you know exactly what is going to happen: https://www.vanityfair.com/news/2018/04/inside-el-faro-the-worst-us-maritime-disaster-in-decades https://www.vanityfair.com/news/2018/04/inside-el-faro-the-w...

dmckeon 8y ago

All else aside, the UI/UX issues here are heart-wrenching. The captain preferred a graphic representation, even though it was 6 to 12 hours out of date, over a more current text-based report/forecast. On two occasions, warnings or new data arrived just a few minutes after a report/forecast was viewed. Crew members watching the Weather Channel had more current/accurate information than the captain. I can easily see how a graphic might be preferred over text, regardless of data currency, and especially during the 95%+ times of smooth sailing, but to see how this situation turned out is troubling.

vpribish 8y ago
I'd like to see the actual BVS UI. Does it indicate how old the data is or just the timestamp the report was generated? This is a detail that bites lots of reports in business.
- marvin 8y ago
  Maybe the people who develop the software don't talk to the end users, or see how they use the software. That would be a classic example of development process gone wrong. Are they, or people sufficiently near to them on the organizational chart, aware that life-or-death decisions are made on the basis of their software? (I'd certainly hope so, plotting weather data definitely qualifies). I'm only conjecturing here; there's no reason to conclude based on the available information that there is a problem here. But it would be a pretty typical situation in an organization that believes software is something you call up a subcontractor to bolt on -- most places are still run like that, although the culture is slowly changing.
  - ethbro 8y ago
    Imho, that's one of the greatest value adds we (software engineers) can bring to the table. The "offloading scope / spec requirements to a PM" trend is a slow motion train wreck. If you, or at least someone else on your team, can't clearly state who's going to be using the code you write, what problem they're using it to solve, and how they're using your code to solve their problem... you need to hire an engineer who can empathize with customers.
  - AnonTemp123 8y ago
    For obvious reasons can't comment too much, but it does, they did, and makes quite clear that it's not for navigational purposes.
    - marvin 8y ago
      Interesting, thanks for the tidbit of inside information :) You'll have to pardon me for asking a stupid question, though -- what other purpose than navigation would a maritime weather report/forecast reasonably have? This seems to me at first look as a protection against liability, rather than an actual clarification of the intended usage of the product. I guess I could imagine a use case such as operations planning, e.g. "will we likely be able to perform work on deck tomorrow?", where the stakes are lower, but I imagine the potential for misunderstanding and incorrect usage of such a service would be dangerously big.
      - AnonTemp123 8y ago
        You're welcome. It's for optimization of future routes -- how do I get from point A to point B while keeping some weather parameters in acceptable range, burning less fuel, and getting there on schedule. And yes, trying not to drive right into a storm. Not quite for trying to maneuver out after you have already driven into one. Of course you can also look at it and figure out of cleaning the deck is a good idea or not. I probably shouldn't comment further, but NTSB's full report is at https://www.ntsb.gov/investigations/AccidentReports/Reports/MAR1701.pdf https://www.ntsb.gov/investigations/AccidentReports/Reports/... and they did not seem to find it too confusing.
- pcunite 8y ago
  It is now StormGeo and looks really nice. No wonder the captain he preferred it. https://vimeo.com/234734651 https://vimeo.com/234734651
hinkley 8y ago
One of the things I’ve learned from doing UI work is that some people absolutely do not picture things in their head. My boss, a generally smart guy, would insist we could squeeze one more thing in somewhere and even though everybody said it wouldn’t fit. So we had to build the stupid thing to show him it wouldn’t work. But until then he needed this thing to happen and neither physics nor his most capable people were going to convince him otherwise. Now, clearly this is a control freak we are dealing with, but also someone in authority who doesn’t trust images or shapes in his head. I had never encountered such a clear case of this happening and it opened my eyes to it happening elsewhere. Some people have low spatial intelligence. You would think becoming a ship captain would cull out a lot of people with that problem but maybe if you’re good enough at other things it doesn’t slow you down that much? Maybe he had to see things to understand them.
Fins 8y ago
From that mentioned Vanity Fair article I sort of got the impression that the captain expected the storm to be similar to what he had experienced when sailing Alaska, not what a proper tropical storm is.

curtis 8y ago

Something that I'd missed in previous articles about the El Faro sinking: > Then, at 0554 , the captain ordered a turn to port to get the wind on the ship’s starboard side, generating a port list. A few minutes later, El Faro lost propulsion --- > Log books indicated no more than 26 inches of oil in the sump in the months before the accident ( a ). With this oil level, the bellmouth would not take in air (lose suction) with an 18-degree list to starboard ( b ), but would lose suction with an equivalent list to port ( c ). They had other serious problems, but loosing propulsion in heavy seas is very serious indeed.

oh_sigh 8y ago
They would have all lived though if they had modern life boats that were enclosed and could be launched from a list
- tgsovlerkhgsel 8y ago
  Are these boats able to reliably withstand such a heavy storm (assuming they don't get smashed against the ship or some rocks/land of course)?
  - lb1lf 8y ago
    Yes. The biggest issue with finding yourself inside one of these in a proper storm is that you’ll get thrown about violently (the lifeboat; you’ll be securely strapped into your seat using a five-point harness, or at least you were in the ones I’ve had evac drills in) for long enough to be as violently ill. I’ve been told by instructors - no guarantee it’s true, that, but imho sounds plausible - that the #1 cause of death in lifeboats or rafts after successfully abandoning ship is dehydration; you sweat profusely, puke repeatedly until your body can’t take it anymore.
sjburt 8y ago
It's interesting that the low oil pressure cutoff (designed to prevent engine damage from running without lubrication) is amongst the things that ultimately doomed the vessel.

mcguire 8y ago

Anyone know if the BVS data has a timestamp? An excellent and very readable document, by the way.

lb1lf 8y ago
-It does; however, the timestamp only shows when the report was compiled; it does not show when the data the report is based on were collected.
- princekolt 8y ago
  As someone who builds software that generates exportable reports, I can say this is one of the cardinal sins of data visualization. Another one is relative timestamps akin to "just now".
  - Treblemaker 8y ago
    I know, right? https://pasteboard.co/Hn35e51.png https://pasteboard.co/Hn35e51.png

ocdtrekkie 8y ago

Definitely impressed by the design work that went into this report. I'd be super curious to hear from people who design these sorts of communications. Usually we think of a well designed pamphlet for marketing purposes, but clear informative presentation here well exceeds what I've traditionally expected from a government report.

njharman 8y ago
It is marketing. 1) they want all to "buy into" their proposed changes. 2) they want all to know how important their work is and why they should remain in the budget.
- tlb 8y ago
  That level of cynicism may apply to some government departments, but agencies like the NTSB are much more grounded. When there's a bad accident on a ship that was more-or-less following procedures, owners and captains of ships really want to know what went wrong and why, so they can avoid it themselves. This is the sort of document that captains will read and discuss with their crew, not all of whom are highly literate in English, so it needs to be accessible and clear. It seems to me that the NTSB did good work here.

mathgenius 8y ago

I'm curious how much this is about "many small and unlikely failures leading to disaster", versus "they sailed into a big storm and were doomed". The usual narrative with complex engineering disasters is the former. I'm wondering, for example, if everything else was going well with the ship, could they have survived such a close encounter with such a big storm? Or, if the engine was kept online, would that have changed anything? It seems to me that this situation went far beyond survivability, but it's hard to tell from the NTSB report.

BrentOzar 8y ago
The fact that the NTSB made 63 different recommendations tells you that it was a complex situation. Their goal is to increase the odds of survival - and 63 recommendations means there were a lot of things that could have turned this situation around just a little. Completely survivable? Maybe not, we’ll never know - but 63 different recommendations means this was a really bad combination of things the crew couldn’t have overcome on their own. Contrast this with a typical plane crash investigation which produces, at worst, a dozen or two recommendations - that tells you a lot. I would have thought the maritime industry would have been better prepared. So sad what the crew was up against.
rdiddly 8y ago
It seems like both... don't sail into a hurricane and there's no problem, right? But even their sailing into it was an interaction of many factors -- the captain's insistence, predicated on preferring the graphically vivid but outdated weather maps, and on disregarding the opinions of his officers, and on each of them independently failing to be sufficiently assertive (or even convinced of the danger themselves) to change his mind. There's also the broken anemometer, but I'm tempted to discount that, because it was obvious the wind was from port, indicating (in any counter-clockwise-rotating storm, i.e. any cyclone in the northern hemisphere) you are headed right into it. Winds should have been from astern if you were trying to cross in front of it. When you're convinced of something, sometimes you can stare right at contrary evidence and disregard it. But who knows, maybe having an accurate wind reading would've jogged some brain cells. If the engine was kept running as you say, that might have helped them, but only if they had used it to turn and get the hell out of there. If the captain insists on using it to continue driving straight into a hurricane, it just makes things worse. And then of course there's all the water they took on, limiting their getaway speed so to speak. In a certain way it seems like the point of no return was the rupture of the fire water inlet. No telling what the ship could've withstood if they had secured the load better and secured the hatches better. Unfortunately all you get with this kind of hindsight is all the things that went wrong, not necessarily what would've had to go right.

wildrhythms 8y ago

This was an incredibly fascinating and shockingly well designed doc. Excellent work from the NTSB putting this together.