North Sea oil rigs spill safety software keeps crashing - no one cares?

[auroraglacialis]

Oh and to finish of a marvelous day with some more enviro news:

In Chevron’s North Sea drill, the oil firm said the spill modelling software usually crashed when left to run for long periods of time, adding that this was typical of standard industry systems.
[....]
Secretary of state Chris Huhne said he was “satisfied that the project is not likely to have a significant effect on the environment”. DECC told the oil giant there was no need for it to provide a full statement detailing exact environmental risk.

Drilling began four days later. This week DECC said in a statement that Chevron had “adequately assessed” the potential impact of a spill.

Chevron bags major North Sea oil contract despite safety software crashes - ComputerworldUK.com

The software is used to model what could be done and where if a spill occurs, so emergency response can be planned in advance. Just weeks before the spill in the gulf of Mexico, the North Sea came close to a similar demise as there was a blowout in a rig - luckily in this case the BOP worked, but the spill modelling would have missed to predict the extent and movement of the spill.

I do not know what worries me more - that the software did not work in this case and the DECC does not see a problem with that, or that the crappy software that fails is typical for standard industry systems!!!

[Isard]

So software that they don't need to use often or for long sometimes fails when left running for long periods of time.


mhm.


And your upset that something that has nothing to do with preventing spills doesn't want to run for 8 or more hours?

[auroraglacialis]

The Oil Spill Information System (OSIS), from BMT Argoss and used by services firm Oil Spill Response, crashed repeatedly when Chevron attempted to run it
[...]
ran the Microsoft Windows-based system for 14 days only because it said “no usable information” could be gleaned from the full run.

So, they do not need to run that program often, but I would not trust a program that cannot do its task because it crashes. Obviously sloppy programming. I mean - no problem - if the program is still in development or was designed only to do 14 days and that was the original intention - but if it is a program that is used in practice AND it was designed to model more then 14 days then it is not really a safe model that one would be sure to suceed in predicting the spread of a spill and the locations of protection boons and skimmers etc.

It's not a critical safety measure, but I think it sort of shows that the motivation to do good work in that field (environmental protection and safety during drilling) is not very high. I am working in environmental science. What would you say if I publish a paper on climate change, modelling the gulf stream and come to the conclusion that it will stop soon, and then someone gets the hands on some information that I exchanged some emails with my project manager talking about the modelling program constantly crashing. Would you not have doubts about the validity of the whole project and publication? I sure would, even though I am all in when it comes to science articels saying that global warming is happening...

[Human No More]

It's not real-time software. Trust me as someone who knows the difference between programming safety-critical and non safety-critical applications, 99.9% of code will NEVER be bugfree. Neither is any application but the smallest and most simple ever 'finished'. Yes, they should dedicate more resources to it, but the way you worded your OP make it sound like it was a realtime system intended for ensuring nothing is actually released or something - pure scaremongering

[auroraglacialis]

Ok, that was not the intention.

To clarify - the program is one that does not run for several days in case of an event, but rather simulates the development of an oil spill under certain conditions (though I cannot say if that simulation runs several hours, days or weeks to produce the model). That simulation covers several days (virtual). The simulation should predict the effect of a period long enough to ensure, that after that time, the spill has ended. This program does not just fail occasionally, but consistently crashes if it tried to model enough days. As a result, they simply accepted a 14 day period as sufficient.

The reason this is worrysome is, that this is also part of a risk assessment for a oil rig. Let's say you want to build an oil rig near the coast and there is a wildlife reservation there. A simulation then can predict, if there is a threat to the reservation due to a spill or if the situation turns out to be manageable. If the spill goes on for a longer time than the model, it is unknown if after these 14 days, the oil will reach the reservation or the situation evolves into one that is harder to manage by emergency protocols. Simply put, you want to know what is happening to the oil in a spill up to the point the spill is over. And what worries me most is, that governments who claim to be spokespersons of the general public and environmental interests do not take such things seriously enough to delay "development" on these grounds.