Northeast Regional 188 - Accident In Philadelphia

Agreed that, on the face of it, this doesn't look computer hacker related. Had it been a terrorist group, they'd be claiming responsibility. Several groups claimed responsibility for Pan Am 103 and I'd assume it would be the same here.

That said, we are certainly in the age where hacking the controls is an issue.

Regarding the second paragraph of EricL's post on the previous page, I can see one possible undesirable effect of using, for example, an Approach indication to protect a 30-mph curve: if the engineer knows the cab signal is going to drop at that location every time, he may begin to assume every time that it's merely the curve restriction that's triggering the signal drop and the cab signal will clear up after the curve has been passed; but if the next signal really is at stop, the engineer won't realize until the cab signal fails to clear up after the curve that it's not just the curve this time--he really needs to be prepared to stop at the next signal. I recall a situation where the most favorable indication on the home signal at a junction (not in cab-signal territory) was Slow Clear, but the distant, having only one signal head, could not display Approach Slow. Consequently, if the home signal was at Slow Clear, you got an Approach; if it was at Stop, you also got an Approach. I often wondered how long it would be before one day the home signal was at Stop because a train was passing through the interlocking and an engineer, expecting a Slow Clear as usual, might not be prepared to stop. And as I recall, the sighting distance at the interlocking wasn't all that great.

ExCon90 wrote:if the engineer knows the cab signal is going to drop at that location every time, he may begin to assume every time that it's merely the curve restriction that's triggering the signal drop and the cab signal will clear up after the curve has been passed;

I agree. Human factors play a role in the design. While solving one problem one can create another problem.

Perhaps in a future version of PTC (beyond simple cab signals) a more detailed display can be provided that shows occupancy on the track ahead. If I recall correctly one of the mockups for PTC had a diagram showing the track ahead as a graph with work zones and other upcoming restrictions clearly noted.

ExCon90 wrote:I often wondered how long it would be before one day the home signal was at Stop because a train was passing through the interlocking and an engineer, expecting a Slow Clear as usual, might not be prepared to stop. And as I recall, the sighting distance at the interlocking wasn't all that great.

I agree that this involves human nature and the law of unintended consequences. But let's not forget, under a positive train control system if the engineer is not prepared to stop, if the engineer does not respond to the next signal (set at stop) and begin reducing speed, the train control system will stop the train. This would seem to be an example of exactly the type of accident PTC is designed to prevent.

To bring this back to on-topic, if ACSES had been activated on the eastward track on which 188 was running the train would have been restricted to 80 mph west of the curve. As soon as the engineer began to accelerate faster than that he would've gotten an audible warning and, if he didn't quickly slow down, a penalty brake.

[quote="Tommy Meehan"]Likewise, if the best aspect that could have been given by the distant signal to Shore (or Shore itself) was Approach Limited, even for non-diverting moves (as is done westbound there and approaching Elizabeth), not slowing down in response to the dropped cab signal would produce the same effect.

Incidentally, SEPTA, with their Rule 562 signaling, uses cab signals to enforce permanent speed restrictions, especially in advance of certain interlockings. As an example, approaching Jenkin on their mainline, the dispatcher has your route lined all the way through the interlocking and beyond to Lansdale. The signals at Jenk South and Jenkin both display Cab Speed. Speed through Jenkin itself is 45mph (#1 track), 30mph (#2 track), and 30mph for diverting moves.
The cab signals on trains will drop from Clear to Approach Medium more than a mile short of Jenk South, then progressively drop to Approach closer to Jenk South, and remain there until clearing Jenkin. All because of the required 30mph within Jenkin.

Regardless, with an Approach in the cab, you better be prepared to stop at the next signal (or drop speed further if the cabs drop to Restricting).

MBTA3247 wrote:While extremely unlikely in this case, remotely hacking an ACS-64 is not a ridiculous idea. The locomotive's diagnostics system constantly reports its status to Amtrak headquarters, which is a possible vector for hacking. There's no reason to believe that Siemens was any smarter about network security than Boeing, medical device makers, or any number of other companies have proven to be.

The Boeing aircraft control system is able to be hacked because Boeing decided to use the same network for aircraft command and control also for passenger Internet access. Running two different pieces of cable/fiber would've avoided this problem.

Does the ACS-64 also receive commands from Amtrak headquarters or is it limited to only one-way communication. I hope the latter. One of the concerns I have about PTC is how hard it is to hack. What prevents someone from deducing the protocol and command set used by receiving and analyzing the transmissions in use?

Silverliner II wrote:Likewise, if the best aspect that could have been given by the distant signal to Shore (or Shore itself) was Approach Limited, even for non-diverting moves (as is done westbound there and approaching Elizabeth), not slowing down in response to the dropped cab signal would produce the same effect.

I read comments from an Amtrak track engineer that the system had put the code in to drop the cab signals at Elizabeth and westbound at Shore because the MAS was high enough that if an engineer didn't slow down for the curves they wouldn't make it around the curve. At Shore eastbound the MAS was 80 mph and that wasn't considered too fast to make the curve. An eastbound rounding Shore at 80 per would produce a very uncomfortable ride but the train would stay on the rails. Westbound the approach speed is 110 mph.

That's interesting about Septa. Their signaling does provide an additional layer of protection. I also understand Septa is one of only three rail companies or agencies currently on schedule to meet the Dec. 2015 deadline to fully implement PTC. (The other two are Amtrak and MetroLink in southern California.)

Unanswered questions of mine -

How much force does it take to advance the throttle on the locomotive type involved in the incident?

Is the throttle notched or infinitely variable?

Don't know if this has been posted in any of the myriad NEC topics or in this thread, but thought I'd add it here as well as in the speed restrictions thread:

New York Times

Brief, fair-use quote:

The Northeast Corridor's Curviest Stretches, and Amtrak's Plans for Safety and Speed

...

Here — by splitting the corridor into mile-long segments of track and ranking each stretch for straightness — we identify the eight sharpest curves on the Northeast Corridor (excluding entrances to stations). Frankford Junction emerges as the most curved stretch between Washington and New York.

...

Other track features besides sharp curves can be problematic. A curve in the tracks in Elizabeth, N.J., is infamous for its low speed limit, even though it’s not sharp — it’s shaped like an elongated letter S.

One of the two sharpest curves on the Northeast Corridor is in New York City: a nearly 90 degree turn on Randalls Island near the Hell Gate Bridge. A curve in Queens, north of the Sunnyside yard, is a bit less curvy, the sixth-most curved.

...

New England hosts some of the most tortuous sections of track. Three of the top eight curves are along the coast in New London County, Conn., where the snaking railroad crosses several rivers over aging bridges and sometimes runs right along the shoreline. The curve in Stonington, Conn., is the sharpest on the corridor. Two other sharp curves (not shown) are in New Haven and Pawtucket, R.I.

There are some really useful maps in this article.

And the "Old Gray Lady's" investigation source: New York Times

The Grey Lady needs to get a grip. There's been 54 homicides in NYC in Q1 2015. That's a Frankford accident every two weeks in terms of body count.

But hey, let's keep making a big deal about the railroad. According to Wikipedia, Amtrak hasn't had 100 deaths in 40 years!!!! Seriously, there's got to be more deaths from freak accidents from walking and chewing gum at the same time.

And of all the crashes and derailments with fatalities Amtrak has had, only in this crash can the blame be assigned to Amtrak at all.

Salem, Essex, Gunpow, Big Bayou Canot, Palo Verde, Silver Spring, Bourbonnais... every single one of them was the fault of events beyond Amtrak's control. Only the amazingly fatality-free Back Bay crash was also Amtrak at fault.

The EGE wrote:And of all the crashes and derailments with fatalities Amtrak has had, only in this crash can the blame be assigned to Amtrak at all.

Salem, Essex, Gunpow, Big Bayou Canot, Palo Verde, Silver Spring, Bourbonnais... every single one of them was the fault of events beyond Amtrak's control. Only the amazingly fatality-free Back Bay crash was also Amtrak at fault.

As well as the Hell Gate head-on, which had one fatality.

Tadman wrote:But hey, let's keep making a big deal about the railroad. According to Wikipedia, Amtrak hasn't had 100 deaths in 40 years!!!! Seriously, there's got to be more deaths from freak accidents from walking and chewing gum at the same time.

I'll wager that within a decade, more people will be killed from texting and walking.

the sarge wrote:

Tadman wrote:But hey, let's keep making a big deal about the railroad. According to Wikipedia, Amtrak hasn't had 100 deaths in 40 years!!!! Seriously, there's got to be more deaths from freak accidents from walking and chewing gum at the same time.

I'll wager that within a decade, more people will be killed from texting and walking.

No bet!