JayUtah
Penultimate Amazing
But then any real evacuation is likely going to be heavily complicated by something, be it bad weather, darkness, lack of time, smoke, blocked escape routes or the reason for the evacuation itself.
Yes, this is the problem in any practical empirical test of evacuation factors. For the Boeing 777 I was a non-qualifying evacuation test subject. That means I was a participant in Boeing's internal evacuation drills that helped them refine their procedure prior to the qualifying drill that established their evacuable capacity under FAR. It's fun; you get to go down the slides. But after doing those drills a half-dozen times, there's no real panic.
Conversely I've been in an ordinary airliner where the cabin lights failed. From that experience I can testify there's a lot of wishful thinking regarding how easy it would be to get out of a cabin under actual conditions. I've also been on a number of seagoing vessels under a number of adverse conditions, including sailing the Great Lakes on a schooner.
But to bring this back to general principles and practice: you're right, every attempt at testing evacuation behavior with real human subjects and achievable apparatus is going to fall short of duplicating the actual circumstances. You have genuinely panicked people. You have smoke. The lights fail. You're disoriented. People are shouting conflicting instructions. You're trying to keep hold of your children.
Which does seem to indicate that using this type of data to pin the starting assumptions of a study, as the Hamburg Report did, is a pretty bad idea.
It's certainly short of ideal. I'm presently in "regulatory hell" on a new product line, so I haven't had enough opportunity to find the paper in which Valenta justifies his use of empirical baselines. The above section is how I feel about the empirical validity of such an approach. It's evidently how you feel. It's how a lot my peer engineers feel. But I don't know if it's how Valenta feels, and I have no clue what his rationale might have been.
In general, people are more comfortable with your models if you can tie them to an empirical foundation. Specifically, regulatory approval that allows simulation models to be used as the verification process will fare better if you can tied the model to an empirical basis, even if it's known to be inadequate. This bites into the constant tension between extremely conscientious engineers who really want to expend their limited resources toward practical safety, and regulators who really demand that you expend your limited resources toward jumping through their hoops.
But I'm not bitter...
IMHO in the MS Estonia case there at the point where the correct coarse of action went from "keep passengers aboard and take rapid efforts to save the ship" to "abandon ship" it was already too late.
I can teach a whole lecture on this. In fact, I have. The decision-making in situations like this follows a predictable pattern. I won't try to reproduce that lecture here. But the salient point of it is what we agree on above: a mandatory evacuation presents real risk of injury and death. Hence you don't order one until you're willing to incur that risk, because you are responsible for the actual consequences should the evacuation later prove unnecessary.
So yes, poor decisions were made here as in many other such tragedies. But don't be too eager to armchair-captain it.