We have to be careful in concluding that the limit for this type of train was actually 50 on this curve. I recall that when the Metroliner's were introduced, into what had been a 90-mph railroad (passenger speeds for GG-1s and heritage cars) with of course reductions on some curves, the Metro's ran at 105 initially, later at 110, in service. On some curves, you would see them restricted to 105 and some 100. But due to stronger brakes and increased clearance at the upper corners, they had generally a 10 mph advantage over the older trains. I assume there may have been special instructions for their operation. I believe they became locomotive-hauled, thus similar to this train except that a newly introduced engine is involved.
The speed limit in the stretch approaching the accident curve was mentioned as being 80 mph. In PC days, the limits for Class A passenger trains (heritage, IIRC) was timetabled there not at 80, but at 70. And then at the accident curve 50, as we see it referred to today. I'm going to suggest that the ability of this train through this area may have been equal moving at 60, to that of a heritage train moving at 50. This would be in regard to excessive sway.
Then, of course, in the NEC, new equipment was always tested for safe operation at a speed 10 miles greater than that intended for scheduled operation. For safe operation, not just that it would go that fast. So this curve could have been safely traversed at 60 by heritage equipment. I'd guess that the accident equipment should have been able to traverse it safely at 70.
I'm not advocating a breaking of the rules. But I thought if would be useful to show what the question is from an engineering standpoint. You have a series of speeds-- 50- 70- 100-- each of which is 40 percent greater that the next lower. It's fairly clear that if the engineman fails to make a one-step speed reduction for a curve, it possibly can still be taken safely by the train.
The significance of the 40% steps in speed is that each such step doubles the overturning moment due to centrifugal force (1.41 x 1.41 = 2). But the speeds for passenger trains on curves are set by what will not upset the comfort of the passengers as they walk down the aisles of the cars. You can arbitrarily set that at a little below half the overturning speed. Of course, these cars are probably more resistant to overturning than their engines. And all engines are not created equal.
So, from an engineering standpoint, the question is, why did this train derail even at 100 miles an hour. Well, of course, the speed was 105, and when the brakes were applied with the train engine and a good part of the cars already in the curve, buff forces developed at the couplers and likely ran in against the engine.
This was an attitude that I learned from the civil engineering department guys of the Pennsylvania. One time there had been a bad electrical accident down in the District, where we were discussing some problems at a proposed overgrade bridge. They asked, why was a boxcar with a ladder ever left parked under wire, and why hadn't the disconnect switch been operated to isolate the siding overhead.
It was then that I learned that it was the Pennsylvania RR that had attacked this problem by having every ladder removed from every last boxcar in America. Well, almost. Did you ever notice that GG-1's had ladders, but they were inside the carbody. They also had a concrete floor one-foot thick, down where all that weight-- 90,000 pounds-- would do more good than just increase traction.
--Frank
We've now heard confirmation that the train was traveling in excess of 100 MPH. 105, 106, 107... take your pick. I heard 106 from the NTSB spokesperson, so let's go with that. Even though the speed limit on the approaching track was 80MPH (pending conditions), the train was still over that limit upon entering the curve, and would have needed to begin its breaking down to 50 
