Inaccurate Stability Calculations Found as Probable Cause in Golden Ray

Strictly speaking the calculation part of the problem was done correctly - the numbers used in the calculations were wrong.

The NTSB determined the probable cause of the capsizing to be the chief officer’s error entering ballast quantities into the stability calculation program, which led to his incorrect determination of the vessel’s stability

Full NTSB report is here:

1 Like

Have found an interesting comment in Maritime Executive. Quote:

Interview of Chief Mate was the funniest interview I ever read. NTSB “determines”? No- NTSB “guesses”. Any lawyer that can not punch holes in the findings should ask their law school for their money back. End quote.

Also “…the operator has switched to a new software system for stability calculations and retrained its officers.” what dovetails with Alias David Boffey thought provoking remarks in latest Splash247 article (comment section) on the issue. . Link to interviews as per gCaptain info

NTSB Docket - Docket Management System

Note: while reading fasten your seatbelts and do not consume any liquids. Latest video on utube worth seeing :

(17) Why Did MV Golden Ray Capsize? NTSB Releases Report | What’s Going on With Shipping? - YouTube


1 Like

Here’s Sal Mercogliano’s post at gCaptain

I’ve haven’t dug into the NTSB report yet but the question as to why did the C/M have the wrong numbers in the computer is a good one.

In the system I’m familiar with the ballast amounts have to be entered manually but evidently on the Golden Ray the tank data can be transferred directly.


It’s often the case fancy automated stuff doesn’t work in practice or quits working and doesn’t get fixed but it certainly would have been worthwhile asking the C/M about it.

The NTSB interview with the C/M was about worthless.

1 Like

On tankers, all the cargo and ballast numbers (and fuel too) are automatically sent to the loading computer. If that broke or stopped working it got fixed ASAP.


That makes sense, not really that fancy to convert the soundings from meters to tons and record the data using a computer. The ship was built in 2017 so most likely it was all still working.

The system was called IMACS and evidently the data was all recovered.

Before I publish my article blasting the ntsb (coming soon) I feel the need to ask an obvious question. How many F’n tanks are in free surface?

And there is zero mention of the term “free surface” in this report. What am I missing?


Goes into more detail on stability then does the final report and includes info about free surface.

It’s # 83 in the docket.

1 Like

I can’t speak for this ship, but on mine anything under 0.5 meters is classified as empty and free surface isn’t considered. It’s because of the longitudinal stringers in the tanks that are that high and drastically reduce the movement of the water.

Also a ballast tank never actually read 0, even though you could dip the tank and it would be dry. So I wouldn’t necessarily put a lot of stock in those numbers being accurate on the low end.


What do you think? Was free surface effect a primary factor?

No I don’t think so. The stability book says something about keeping free surface to a minimum as a general rule but the free surface correction is included in the final computed GM.

1 Like

Page 24 and page 25 of the NTSB report says that the vessel was also out of compliance on the previous voyage legs: Freeport / Jacksonville / Brunswick

Maybe only in compliance when the C/M took on ballast for Hurricane Dorian where 1500 tons (according to IMAC data) was added but then pumped out again after.


There doesn’t seem to be much information in the report about the draft restrictions in Jacksonville and Brunswick. Apparently both ports had about the same depth restrictions but not much about time restrictions wrt tide.

The report lists TPC as 56.10 MT/cm (equivalent of tons per inch immersion) so 1500 tons of ballast would be less then 0.3 meters increase in draft.

EDIT: The fact that the C/M added 1500 tons of ballast for hurricane Dorian would seem to indicate that he understood the ship had insufficient ballast.

Assuming Sal Mercogliano is correct that the C/M entered the ballast amounts manually to produce reports that made the ship appear to be in compliance with righting arm criteria than he was playing with fire.

That extra ballast is not just for wind heel.

Thoughts sent to me from the master of a RoRo:

When I’m tight on stability the feeling when I raise the ramp is the first thing I look for. Then I see how we feel when the tugs pull us. There is no way it didn’t feel like shit when the tugs pulled them

There’s that word @Kennebec_Captain used again: Feel

1 Like

Container ship guy here. When I feel her take a pronounced heel in a turn departing port or she is wobbling back and forth as the quartermaster is using rudder, I start eyeballing the Chief Mate and want to know when the last time manual soundings were taken.

Feel is everything.


It’s definitely true about the ramp coming up and when tugs start to pull. I’m not sure how well calibrated that feel is.

In the case of the Hoieg Oska the calculated GM was 1.46 meters, the actual GM was never determined but it was estimated to be 0.7 meters In that case the master told the mate the ship didn’t feel right and there was a big heel when the ramp was raised.

In the case of the Golden Ray I haven’t looked at the reports enough yet but the MSC report seems to be saying the GM was 1.76 meters. In one place in the report is mentioned a GM of 0.45 but I’m not sure if that’s initial GM.

The ship would have been OK at less then 5 degrees but once it goes past 10 or 15 degrees however not much righting arm.


The other issue being the flooding from the pilot door being open.

Here are the numbers from the NTSB report:


At low speed the ship didn’t heel much. Doesn’t look like much warning before the final turn.



The root cause the NTSB found is that stability data entered was incorrect and give the reason the C/M made errors was lack of training.

Sal Mercogliano is correct that the issue with the findings is that the NTSB didn’t ask why the C/M made errors:

For example this is strange:


This is done frequently when there is not a good pre-stow or if the pre-stow is sent too late but in this case the charter provided a good stow plan that gave the weights of each hold and deck, which is exactly what is needed. When did the C/M get the stow plan? What info was the C/M using that he had to estimate the weights?

The other unanswered question is why didn’t the C/M use the IMAC system to enter the data?

I didn’t see anything about work/rest hours in the report.

The original schedule was Freeport - Brunswick - Jacksonville but the charterer switched to JAX first then Brunswick. That would have made the calculated stability for both ports incorrect and require it be redone.

Another issue that wasn’t addressed was the draft restrictions. Was it possible for ship to meet the conflicting requirement for sufficient ballast and max draft limits? If so at what tide?

This is interesting - the vessel’s IMAC system had a feature that could estimate GM by doing the equivalent of an inclining experiment by measuring the change in list when moving ballast:


GM of 0.8 meters. Don’t really know but that GM makes more sense to me then the 1.76 meters the NTSB was using. No mention in the report of how this discrepancy was resolved.

Are you familiar with this feature @New3M ?


I’m not sure which feature you are asking about - the calculation of GM?

The stability of a tanker is much different than that of a car carrier. I’ve never worked on one or had to guesstimate how much each car I’m loading and discharging weighs for my stability.

We use CargoMax for all of our calculations, and yes, at any moment you can force the program to run a calculation to calculate your “actual” GM. The program calculates an estimated GM each time it updates (every 15 seconds usually), but in order to get a full GM calc you need to run the calculation, which takes a few minutes typically. Usually I only do this at the end of a port stay, but I’m not dealing with as close a margin as they were here.

I also never put a lot of stock in the calculations for the list via CargoMax - you can play with ballast numbers to make the list be 0, but in actuality it was usually quite different.

1 Like

Not the calculations by just grinding the numbers but the quasi-inclining experiment feature where the list is corrected by shifting ballast and the program uses the change in list and the amount of ballast used to calculate GM,

Only needed 8 tons of ballast to take off half a degree of list, sounds like a very tender condition.

Well, that’s something that our program does automatically each time it updates all online tanks. Every cargo, ballast, and fuel tank has a sensor in it that updates every 15, 30, or 60 seconds depending on what you select. And each time it does, it recalculates everything - trim, drafts, heel, bending and shear moments, GMt and GM margin, among other things.

I don’t know anything about this program, but I would think it would do the same. Maybe calculating different moments based on being a car carrier and not a tanker, but the broad strokes should be very similar I would think.

The IMAC system calculated the same way we are familiar with but it could also calculate GM a second, independent way in addition - sort of an Inclining test.

The IMACS also had a GM measurement feature that, if used, automatically transferred ballast to heel the vessel up to 1° to either side and measured the quantity of ballast transferred to calculate the ship’s GM.


The GM position is determined by moving weights transversely to produce a known overturning moment in the range of 1-4 degrees if possible.[2] Knowing the restoring properties (buoyancy) of the vessel from its dimensions and floating position and measuring the equilibrium angle of the weighted vessel, the GM can be calculated.

Similar to estimated the GM when the ramp is lifted. A rough estimate can be made by how much the ship rolls when the ramp comes off the pier,