MS Estonia: what the JAIC might have (actually) gotten wrong & conspiracy theories
Long time reader, first time poster. Anyway, with the MS Estonia investigation being reopened alongside a bunch of CT's doing their own "investigation," I figured now would be a good time to bring this stuff up.
First, I would consider myself a well educated layman on this and many other engineering topics. (Meaning if an actual expert disagrees with me I should listen to them.) I've read a ton of NTSB reports since they're interesting to me, and a whole bunch of technical accounts of damaged WWII warships sinking (or not sinking.) (Captain_Swoop, still have the link to a WWII era document on progressive flooding you mentioned in that 5-part trainwreck of a thread? Sounds like interesting reading. I actually have a hardcopy of the the damage control handbook you posted a link to.)
And to me, it's abundantly obvious how the Estonia sank - systematically under engineered bow door attachments on Baltic RO-RO ferries combined with a poor crew (and perhaps a touch of bad luck, and that the visor couldn't be seen from the bridge) resulted in a disaster that was eventually going to happen to some ferry sooner or later. The MS Estonia just drew the short end of the straw. And a RO-RO ferry capsizing is something that tends to happen on a fairly regular basis. As far as the basic series of events the JAIC is correct. And I can certainty imagine the magnitude of the free-surface effects 2000 tons of water on a car deck that size. But...
About the details. (A change to chat with engineers yay!) What time did the first attachment on the bow visor break, when did it come off, what was the history of the angle of heel? (I have a suspicion that the JAIC may have interpreted evidence to put the time of the bow visor going over the side at the latest supportable time, with the perfectly mundane motive of making the crew look less incompetent. Boring stuff that happens in real life.) A paper from 2002 (sorry read it a while ago, no link handy) said that nobody's model accounted for both the ship sinking and a steady high list angle for a short period of time. (Did seem a little simplified in hindsight.)
Of course, that was in 2002, and since then advances in computer hardware and money being spent resulted in a a pair of studies going on at roughly the same period of 2006-2008. First in the Hamburg Ship Model Basin report (which is probably the wrong attribution, "Hamburg Report" from here on out, not allowed to post a link it turns out), they conclude that the bow visor went overboard at 1:00-1:02, based on their "evacuation study" (more on that later) concluding that in order for the crew in the ECR to have made it up the escape shaft in the funnel in time the clock there was off by 12-13 minutes. (To me, that seems to be a pretty big assumption to make, and the evacuation study had better be solid...)
They did what I can only assume is a decent numerical simulation using the chosen inputs, and note the importance of the downflooding through ventilators. (I would assume the JAIC in what they did took that into account, but I also have a hard time buying that they did and their report where talking about flooding through the windows doesn't also include something along the lines of "...along with flooding through ventilation..." They also did a simulation of the initial flooding after the bow visor jumped ship and noted that a starboard turn would have rapidly capsized the ship, and that the difference between "ship survives and returns to port" and "ship rolls right over rapidly" is as little as 30 seconds. However, they seem to assume everywhere a high degree of certainty in their wording and conclusions that seems unjustified considering the uncertainty. I don't know if that matters or not, but it does strike me the wrong way.
Their list history seems to work out with a list going from 0 degrees to a steady ~30 degrees over ~7-9 minutes (Huh? Wasn't there a "big heel," which I suppose could have been the result of wave action, but still...), and then staying steady at 30 degrees for nearly 10 minutes before continuing to roll over.
Contunuing to the (had better be good) evacuation model it seems on page 81/89 (depending on report or PDF page number) we go boldly extrapolating past the realm of data. It bases almost everything on the coefficient of friction between shoes and deck, and seems to be a little... lacking in empirical data. Also it seems to somewhat ignore that many survivors actually said walking without support was impossible, but on a second reading of the paper I'm no longer sure. It also doesn't consider the whole "bulkheads become deck" thing. As for that all I could find (and I dug pretty deep into references) was (from a newer paper) a brief description of a crude but probably reasonable effective experiment of "walking on floor and wall" done by somebody else, with only two ~20 female volunteers for data (I'm assuming friends or members of the research team), although not taking into account of the "bulkhead becomes overhead" issue where the ceiling gets lower (depending on width of corridor.), and that reference only went to a dead Youtube link. (The JAIC determined that escape became impossible past ~45-50 degrees due to corridor geometry.)
Running it, and comparing the number of people who make it topside (which is correct, as this shouldn't take into account those who did but drowned), they get... twice as many people making it topside as who should. After increasing the list they calculated by 2.5 degrees (okay), they get about the right figure, but only a little over a third of the actual survivors made it topside. (About that "model had better be good" thing earlier...) Then they calculate the escape of the crew in the ECR, assuming they have to make it topside before the list exceeds 45 degrees (ignoring survivors who saw them emerge when it was closer to 90, and that an enclosed escape stairway would probably allow relatively easy walking/crawling along the wall and plenty of handholds).
(Anybody who's been aboard a sailing ship at a high angle of heel able to provide insight? I'd assume those working on deck wouldn't be trapped against the lee rail, ((IIRC) some of the lines are handled from there???) even if (apparently) the heel is 40-45, or maybe even 55 degrees.)
Then that still don't result in them making it topside before the assumed list angle makes it impossible. Which was the whole basis for the assumed time of visor loss. Then they conclude that (rather than possibly faulty assumptions) than implies the visor fell off even earlier. (So go rerun your model then if you're so confident...) So the whole thing is based on circular reasoning that doesn't even meet up and the ends.
The SSPA report I'm inclined to consider more reliable both due to cross-checking and validation with model tests (which having a large budget allows) and honestly due to expressing the proper degree of uncertainty. ("Most probably sinking scenario," for example.) Their series of model runs rule out underwater damage as the cause (although the project report that they do that in no longer seems to be available.). Also their test basin results hint as the ramp moving up and down under wave action rather than remaining fully open. (Which I'd assume is something that examination of the soon to be recovered bow ramp could reveal, so check for that and incorporate as needed. Also - what if the bow ramp never opened fully, and was still closed by about a meter? It would then trap water on the car deck, or wave action could lead to temporary retaining of water. Good thing the bow ramp is being raised!) Their numerical model runs result in a rapid increase of list to ~40-45 degrees, slowly increasing before eventually increasing rapidly.
That said, all their numerical modelling assumed all internal doors were open. My ignorant self would assume a cabin with the door closed would take a few minutes to flood. (something like maybe 2-3 perhaps?) (Captain_Swoop am I anything close to right about this?) A half sized model pf a cabin constructed identically (except at half thickness etc.) would actually fit in the model basin they used, and might provide some useful data. (Alternately a full-scale and a lake would be possible.) (Alternately again a full sized copy of one of the smaller cabin could well fit in the model basin.)
Then they go to a very big (about 30 feet long IIRC, said the model basin was big) model with a complete and identical (although simplified) interior. It sinks (in the late stages) pretty much identically to survivor accounts. There's a picture in that section of the ship model as it's going beneath the surface that would be very useful in showing why the ramp closed due to gravity - seeing the angle the ship assumed it's blatantly obvious simply looking at the picture that it'd be almost certain to.
However, the list history quickly goes to 40 degrees, and then slowly increases at a constant rate to 100 degrees before rapidly completing the process, which completely contradicts the numerical model. And finally the whole study from the early stages went with the list slowly increasing to a full 20 degrees due to water entering the car deck around the partially opened bow visor after the first failure but before it separated entirety, before the first big heel.
Honestly, I think a major part of the problem is that the original investigation simply didn't provide enough data points to even show what shape the curve should be (which could in theory be provided by a single survivor sans-values; "it heeled over a bunch, stayed maybe the same for a bit then heeled fast again" type statements in large quantities would be invaluable.) Just about any shape of the list-time curve can be supported with what we have currently by determining which people to use for data points. By now witness memory is very contaminated, but we also know a bit more about memory now, and hopefully now they're interviewing everybody the errors will average out and we'll finally have either a decent list-time history, or at least know which shape the curve on the graph should take! (For numerical modelling.)
Also - conspiracy theories!
A few thoughts first - perhaps the reason most of the original dive footage wasn't released was since the parts showing bodies were cut out? Nowadays it seems they're releasing all of it with gray circles over the bodies.
Of course when people want a new investigation they can 'trust' they mean 'one that gives the results I want.'
I'm somewhat alarmed by how high our old friend Hewia places in the google search results.
Perhaps a few of the crew witnesses were "fudging" times in interviews in order to make themselves look better?
I don't see why recovery of the wreck would have been impossible. Enormous engineering challenge and incredibly expensive yes, and still not recovering all bodies and likely destroying some in the process, but not completely impossible, just impractical and not worthwhile.
Anyway, first I'm wondering about the history of conspiracy theories about the Estonia. Who invented what when for example. Also what all has Jutta Rebe (sp) done to disgrace herself? With the recording in the pilot logbooks, I can certainly find it plausible that on the flight back to shore some other non-pilot member of the crew on at least one helicopter had the idea to ask name and DOB of those who were in good enough shape to answer, and the pilot logbook was the only paper handy (simply written all over the page) (there would have easily been enough time for that, and it's not like, say, the winch operator has anything better to do) but it's obvious she's fabricating seeing the captain and senior officer's names on one. (Also - even if she did see one, would she even be able to recognize the name? I doubt she can read Swedish or Finnish or Estonian, and I don't know how close German is.)
Also wouldn't the senor officers be, well, older? On account of the time needed to make that rank? Older people on the Estonia fared very poorly after the sinking, so for all we know some of the senior crew were able to make it topside, then drowned outside quite quickly, or being older by the time the bridge was abandoned were unable to make it out due to the list. There's no evidence the captain ever made it to the bridge, for example.
Finally, we might as well have some fun. Surely we can come up with some good conspiracy theories! Good, of course, measured by what would make the most "best"/most entertaining awful cheesy low-budget spy thriller, with tons of plot holes and little relationship to reality.
Long time reader, first time poster. Anyway, with the MS Estonia investigation being reopened alongside a bunch of CT's doing their own "investigation," I figured now would be a good time to bring this stuff up.
First, I would consider myself a well educated layman on this and many other engineering topics. (Meaning if an actual expert disagrees with me I should listen to them.) I've read a ton of NTSB reports since they're interesting to me, and a whole bunch of technical accounts of damaged WWII warships sinking (or not sinking.) (Captain_Swoop, still have the link to a WWII era document on progressive flooding you mentioned in that 5-part trainwreck of a thread? Sounds like interesting reading. I actually have a hardcopy of the the damage control handbook you posted a link to.)
And to me, it's abundantly obvious how the Estonia sank - systematically under engineered bow door attachments on Baltic RO-RO ferries combined with a poor crew (and perhaps a touch of bad luck, and that the visor couldn't be seen from the bridge) resulted in a disaster that was eventually going to happen to some ferry sooner or later. The MS Estonia just drew the short end of the straw. And a RO-RO ferry capsizing is something that tends to happen on a fairly regular basis. As far as the basic series of events the JAIC is correct. And I can certainty imagine the magnitude of the free-surface effects 2000 tons of water on a car deck that size. But...
About the details. (A change to chat with engineers yay!) What time did the first attachment on the bow visor break, when did it come off, what was the history of the angle of heel? (I have a suspicion that the JAIC may have interpreted evidence to put the time of the bow visor going over the side at the latest supportable time, with the perfectly mundane motive of making the crew look less incompetent. Boring stuff that happens in real life.) A paper from 2002 (sorry read it a while ago, no link handy) said that nobody's model accounted for both the ship sinking and a steady high list angle for a short period of time. (Did seem a little simplified in hindsight.)
Of course, that was in 2002, and since then advances in computer hardware and money being spent resulted in a a pair of studies going on at roughly the same period of 2006-2008. First in the Hamburg Ship Model Basin report (which is probably the wrong attribution, "Hamburg Report" from here on out, not allowed to post a link it turns out), they conclude that the bow visor went overboard at 1:00-1:02, based on their "evacuation study" (more on that later) concluding that in order for the crew in the ECR to have made it up the escape shaft in the funnel in time the clock there was off by 12-13 minutes. (To me, that seems to be a pretty big assumption to make, and the evacuation study had better be solid...)
They did what I can only assume is a decent numerical simulation using the chosen inputs, and note the importance of the downflooding through ventilators. (I would assume the JAIC in what they did took that into account, but I also have a hard time buying that they did and their report where talking about flooding through the windows doesn't also include something along the lines of "...along with flooding through ventilation..." They also did a simulation of the initial flooding after the bow visor jumped ship and noted that a starboard turn would have rapidly capsized the ship, and that the difference between "ship survives and returns to port" and "ship rolls right over rapidly" is as little as 30 seconds. However, they seem to assume everywhere a high degree of certainty in their wording and conclusions that seems unjustified considering the uncertainty. I don't know if that matters or not, but it does strike me the wrong way.
Their list history seems to work out with a list going from 0 degrees to a steady ~30 degrees over ~7-9 minutes (Huh? Wasn't there a "big heel," which I suppose could have been the result of wave action, but still...), and then staying steady at 30 degrees for nearly 10 minutes before continuing to roll over.
Contunuing to the (had better be good) evacuation model it seems on page 81/89 (depending on report or PDF page number) we go boldly extrapolating past the realm of data. It bases almost everything on the coefficient of friction between shoes and deck, and seems to be a little... lacking in empirical data. Also it seems to somewhat ignore that many survivors actually said walking without support was impossible, but on a second reading of the paper I'm no longer sure. It also doesn't consider the whole "bulkheads become deck" thing. As for that all I could find (and I dug pretty deep into references) was (from a newer paper) a brief description of a crude but probably reasonable effective experiment of "walking on floor and wall" done by somebody else, with only two ~20 female volunteers for data (I'm assuming friends or members of the research team), although not taking into account of the "bulkhead becomes overhead" issue where the ceiling gets lower (depending on width of corridor.), and that reference only went to a dead Youtube link. (The JAIC determined that escape became impossible past ~45-50 degrees due to corridor geometry.)
Running it, and comparing the number of people who make it topside (which is correct, as this shouldn't take into account those who did but drowned), they get... twice as many people making it topside as who should. After increasing the list they calculated by 2.5 degrees (okay), they get about the right figure, but only a little over a third of the actual survivors made it topside. (About that "model had better be good" thing earlier...) Then they calculate the escape of the crew in the ECR, assuming they have to make it topside before the list exceeds 45 degrees (ignoring survivors who saw them emerge when it was closer to 90, and that an enclosed escape stairway would probably allow relatively easy walking/crawling along the wall and plenty of handholds).
(Anybody who's been aboard a sailing ship at a high angle of heel able to provide insight? I'd assume those working on deck wouldn't be trapped against the lee rail, ((IIRC) some of the lines are handled from there???) even if (apparently) the heel is 40-45, or maybe even 55 degrees.)
Then that still don't result in them making it topside before the assumed list angle makes it impossible. Which was the whole basis for the assumed time of visor loss. Then they conclude that (rather than possibly faulty assumptions) than implies the visor fell off even earlier. (So go rerun your model then if you're so confident...) So the whole thing is based on circular reasoning that doesn't even meet up and the ends.
The SSPA report I'm inclined to consider more reliable both due to cross-checking and validation with model tests (which having a large budget allows) and honestly due to expressing the proper degree of uncertainty. ("Most probably sinking scenario," for example.) Their series of model runs rule out underwater damage as the cause (although the project report that they do that in no longer seems to be available.). Also their test basin results hint as the ramp moving up and down under wave action rather than remaining fully open. (Which I'd assume is something that examination of the soon to be recovered bow ramp could reveal, so check for that and incorporate as needed. Also - what if the bow ramp never opened fully, and was still closed by about a meter? It would then trap water on the car deck, or wave action could lead to temporary retaining of water. Good thing the bow ramp is being raised!) Their numerical model runs result in a rapid increase of list to ~40-45 degrees, slowly increasing before eventually increasing rapidly.
That said, all their numerical modelling assumed all internal doors were open. My ignorant self would assume a cabin with the door closed would take a few minutes to flood. (something like maybe 2-3 perhaps?) (Captain_Swoop am I anything close to right about this?) A half sized model pf a cabin constructed identically (except at half thickness etc.) would actually fit in the model basin they used, and might provide some useful data. (Alternately a full-scale and a lake would be possible.) (Alternately again a full sized copy of one of the smaller cabin could well fit in the model basin.)
Then they go to a very big (about 30 feet long IIRC, said the model basin was big) model with a complete and identical (although simplified) interior. It sinks (in the late stages) pretty much identically to survivor accounts. There's a picture in that section of the ship model as it's going beneath the surface that would be very useful in showing why the ramp closed due to gravity - seeing the angle the ship assumed it's blatantly obvious simply looking at the picture that it'd be almost certain to.
However, the list history quickly goes to 40 degrees, and then slowly increases at a constant rate to 100 degrees before rapidly completing the process, which completely contradicts the numerical model. And finally the whole study from the early stages went with the list slowly increasing to a full 20 degrees due to water entering the car deck around the partially opened bow visor after the first failure but before it separated entirety, before the first big heel.
Honestly, I think a major part of the problem is that the original investigation simply didn't provide enough data points to even show what shape the curve should be (which could in theory be provided by a single survivor sans-values; "it heeled over a bunch, stayed maybe the same for a bit then heeled fast again" type statements in large quantities would be invaluable.) Just about any shape of the list-time curve can be supported with what we have currently by determining which people to use for data points. By now witness memory is very contaminated, but we also know a bit more about memory now, and hopefully now they're interviewing everybody the errors will average out and we'll finally have either a decent list-time history, or at least know which shape the curve on the graph should take! (For numerical modelling.)
Also - conspiracy theories!
A few thoughts first - perhaps the reason most of the original dive footage wasn't released was since the parts showing bodies were cut out? Nowadays it seems they're releasing all of it with gray circles over the bodies.
Of course when people want a new investigation they can 'trust' they mean 'one that gives the results I want.'
I'm somewhat alarmed by how high our old friend Hewia places in the google search results.
Perhaps a few of the crew witnesses were "fudging" times in interviews in order to make themselves look better?
I don't see why recovery of the wreck would have been impossible. Enormous engineering challenge and incredibly expensive yes, and still not recovering all bodies and likely destroying some in the process, but not completely impossible, just impractical and not worthwhile.
Anyway, first I'm wondering about the history of conspiracy theories about the Estonia. Who invented what when for example. Also what all has Jutta Rebe (sp) done to disgrace herself? With the recording in the pilot logbooks, I can certainly find it plausible that on the flight back to shore some other non-pilot member of the crew on at least one helicopter had the idea to ask name and DOB of those who were in good enough shape to answer, and the pilot logbook was the only paper handy (simply written all over the page) (there would have easily been enough time for that, and it's not like, say, the winch operator has anything better to do) but it's obvious she's fabricating seeing the captain and senior officer's names on one. (Also - even if she did see one, would she even be able to recognize the name? I doubt she can read Swedish or Finnish or Estonian, and I don't know how close German is.)
Also wouldn't the senor officers be, well, older? On account of the time needed to make that rank? Older people on the Estonia fared very poorly after the sinking, so for all we know some of the senior crew were able to make it topside, then drowned outside quite quickly, or being older by the time the bridge was abandoned were unable to make it out due to the list. There's no evidence the captain ever made it to the bridge, for example.
Finally, we might as well have some fun. Surely we can come up with some good conspiracy theories! Good, of course, measured by what would make the most "best"/most entertaining awful cheesy low-budget spy thriller, with tons of plot holes and little relationship to reality.
