Amtrak 188 wreck: NTSB responds; Schanoes rebuts

David Schanoes’ personal analysis of the Amtrak train No. 188 derailment at Frankford Junction curve on the Northeast Corridor last year elicited a response from National Transportation Safety Board Director Robert J. Hall, P.E. We publish it in full, followed by Schanoes’s response:

I would like the opportunity to address some of the comments made by David Schanoes in the Railway Age article, “The short, the long, the skinny, and the fat” regarding the National Transportation Safety Board (NTSB) investigation and findings regarding the derailment of Amtrak train 188 in Philadelphia, Pennsylvania, on May 12, 2015.

The article states, “NTSB is not going to tell you anything 12 months from the incident that you won’t already know within a day or two, if you do your work, and if you don’t abdicate your responsibility.”

A comprehensive accident investigation cannot be conducted hastily. It is an exhaustive process that involves developing and supporting the most plausible factors contributing to the accident. Also as challenging is the need to rule out factors that we know can lead to an accident. Consider the effort needed to eliminate the following variables that the NTSB has previously identified as contributing to major transportation accidents: operator fatigue (due to irregular and unpredictable crew scheduling, obstructive sleep apnea, circadian disruption); medical conditions (including seizures, diabetes, and color blindness); use of prescription and nonprescription medications; illicit drugs; alcohol; inadequate training and testing of operating and signal rules; poor signal conspicuity; train dispatcher errors; and distraction due to text messaging and cell phone calls. It takes time and great effort to analyze these variables to assess if they played a role in an accident. From the NTSB’s perspective, any conclusion about the cause of the Amtrak accident that was formulated just days after the accident would be purely speculative, incomplete, and lacking a thorough analysis of relevant information.

The article states, “Call it loss of situational awareness, if you want. I like calling it what it is—failure to properly control the speed of the train, no matter what.”

This statement is a simple explanation of what happened. Indeed, event recorder data showed that the train entered the Frankford Junction curve, with a maximum authorized speed of 50 mph, traveling at 106 mph. However, presenting only what happened does not explain why the engineer was operating his train twice the safe speed. Corrective actions can best be taken when an investigation can get down to the root cause of the accident to determine why something went so terribly wrong. The NTSB report documents post-accident actions by Amtrak and the Federal Railroad Administration and makes several safety recommendations to prevent similar accidents.

The article states, “We’ve gone from speculation to assertion to confirmation, and we’ve done all this based on … evidence? Not exactly. Actually, not even close.”

There are always some facts or questions that cannot be fully answered in an accident investigation, including this Amtrak accident. Nonetheless, as evidenced by the NTSB public docket, investigators were able to analyze a tremendous amount of information. This includes the engineer’s training records, rules exams, efficiency testing, performance evaluations, time sheets, medical records from Amtrak and the engineer’s private physician, as well as additional medical evaluations he took at the request of the NTSB. Beyond that, investigators reviewed scientific studies—including research in the areas of situational awareness, distraction and information processing, and prospective memory—to support their analysis. The NTSB determines the probable cause of an accident, and, as necessary, makes recommendations aimed at preventing future accidents. In the Amtrak 188 investigation, we made 14 new recommendations.

The article stated, “Dr. Jenner claims that those (radio) conversations last for 6 minutes, with the final conversation at 21:19:13. Actually, Dr. Jenner is mistaken. The final conversation between SEPTA (Southeastern Pennsylvania Transportation Authority) train 769 and the train dispatcher … took place at 21:18:01. A subsequent remark, made a minute later-unidentified in origin … produced the following: Second Person [laughs]: ‘Yeah, we got rocked.’”

The last radio broadcast, as you pointed out, occurred at 21:19:13. There is too much static from that radio broadcast for it to be understood. As a result, the Amtrak engineer would not have known if the transmission originated from the dispatcher or from someone else. Nonetheless, the response was clearly made by the SEPTA engineer, and it is consistent with the SEPTA emergency situation—“We got rocked”—and which makes it likely that the Amtrak engineer was interested in what the SEPTA engineer had to say. Nonetheless, even without that last radio transmission, the SEPTA communication with the dispatcher lasted about 5 minutes, still a significant amount of time to have one’s attention diverted while still attempting to focus on his own train operations.

The article states, “In fact, the event recorder shows that he operated the train at the proper speed … until, realizing his error, he initiates an emergency brake application.”

While listening to the communications between the SEPTA engineer and the dispatcher, the Amtrak engineer did operate his train at or near track speed. This is not surprising. People who are distracted, for instance, those who drive their cars while on their cell phones or texting, are often able to maintain a legal speed and stay in their lanes for a period of time. Nonetheless, those who are distracted when using cell phones and texting are subject to “cognitive tunneling”—in short, an inattentional blindness phenomenon in which the observer is too focused on instrumentation, task at hand, internal thought, and the like, and not on the present environment. Similarly, though the Amtrak engineer was not on his cell phone, his attention on the SEPTA train caused him to lose focus on other tasks, and not on the present environment. Moreover, distraction research, dating back more than 50 years, has clearly demonstrated its degrading effects on a person’s recall and ability to process information fully. Hence, the report’s conclusion: “The Amtrak engineer accelerated the train to 106 mph without slowing the train for the curve at Frankford Junction, due to his loss of situational awareness, likely because his attention was diverted to the emergency situation with the SEPTA train.” The article presents no alternative explanation.

We share an interest in making the transportation systems in the United States among the safest in the world. In the spirit of mutual understanding, I invite you and your staff to come and visit the NTSB, meet with our investigators and scientists, tour our laboratories, and talk with our Board members. I hope you will accept my invitation and come with an open mind to learn about how the NTSB makes transportation safer for everyone.

David Schanoes responds:

Thanks to Mr. Hall for taking the effort to respond to some of the issues I raised, and in a very timely manner.

Needless to say, Mr. Hall and I disagree. Mr. Hall claims that the 12-month investigation period was essential and unavoidable given the number of factors NTSB had to consider: “It is an exhaustive process that involves developing and supporting the most plausible factions … Also challenging is the need to rule out factors that we know can lead to an accident …”, all of which leads Mr. Hall to conclude that: “From the NTSB perspective, any conclusion about the cause of the Amtrak accident that was formulated just days after the accident would be purely speculative, incomplete, and lacking a thorough analysis of relevant information.”

NTSB only decides to analyze a certain number and types of railroad accident. For example, NTSB has not investigated the March 2016 derailment of Altamont Corridor Express train #10. Does that mean that an investigation conducted without the NTSB, by the operating officers of that railroad, and without access to NTSB’s resources, will be purely speculative and incomplete? Of course not.

Let’s look at the transport of crude oil by rail. NTSB has participated in several investigations of those derailments, but decided not to participate in last July’s derailment of a BNSF CBR train in Montana, nor the recent derailment of a UP train carrying Bakken crude in Oregon. Does that mean the investigations of those derailments by those railroads’ operating officers are speculative, lacking a thorough analysis? Of course not.

Railroad operating officers are trained to access and examine the same data NTSB utilizes. And they are trained to reach their conclusion quickly so that the proper preventive measures can be determined and instituted. Were those determinations “hasty,” “speculative,” or “incomplete”? The decades-long trend in improved safe rail operations proves that those determinations were based on data, thorough, and detailed. Have the industry’s efforts been perfect? No. Have we required prodding from the regulatory agency? Yes. But has the industry conducted thorough investigations determining cause and identifying preventive measures? Yes it has.

But there’s a bit more to this than just professional pride. Contrary to what Mr. Hall implies, it is NTSB itself that introduces speculation and then determines, and enshrines that speculation as cause.

In the case of Amtrak #188, it is NTSB that speculates, without any direct evidence, that the locomotive engineer was “likely distracted” by the radio conversation between the Amtrak train dispatcher and SEPTA train #769. It is NTSB that speculates, without any direct evidence, that the locomotive engineer lost track of where he was, due to his concern with the condition of the engineer of #769. The locomotive engineer himself claims no memory of the last several minutes of train operation. However, the data from the event recorder shows he is in control of the train, is operating the throttle, and is responding properly to the alertness device during that period of time. The data points to no distraction due to railroad radio use.

Moreover, the notion that railroad radio transmissions are a “distraction” is directly contrary to the evidence of thousands of train movements daily in the U.S., where radio transmissions providing information regarding conditions affecting train movements are integral to safe train operations. Proper use of railroad radios is anything but distracting to locomotive engineers responsible for safe train movement. Comparing proper railroad radio usage to an automobile driver’s use of a cell phone to text, as Mr. Hall does, misses the essential difference between railroad radios and cell phones in an automobile. That difference is that railroad radios are part of the safe operating environment of the railroad. Locomotive engineers are trained on proper use of, and attention to, the railroad radio in order to achieve and satisfy the primary obligation—to properly control the speed of the train. A cell phone is not integral to, and is in fact, a negative factor in safely controlling an automobile.

Consider, for a minute, the direct implications of the NTSB’s argument—that an instrument integral to safe train operations distracted the locomotive engineer and likely contributed to his loss of situational awareness. What then is a railroad to do? “Train” its engineers to not be “distracted” by the necessity of paying attention to radio transmissions? “Train” its engineers to “multitask?” That, to use an unkind word, would be nonsense. And, just to be totally unkind and totally accurate, total nonsense.

We do not train locomotive engineers to “multitask.” We train locomotive engineers to execute a single task, at all times, and under all conditions: to properly control the speed of the train. One, or the NTSB, might as well recommend that railroads cease issuing summary bulletin orders, reiterating speed restrictions and other information affecting the safe movement of trains, because it’s possible a locomotive engineer at the controls of a moving train might read the order and while reading the bulletin order, miss a braking point and possibly derail the train.

Mr. Hall points out that I do not offer an alternative cause for the locomotive engineer’s “loss of situational awareness.” Indeed, I don’t, because I know the cause for the derailment: the locomotive engineer’s failure to properly control the speed of the train under all circumstances. What counts is what we know, not what we speculate.

I would also like to thank Mr. Hall for pointing out that NTSB has issued 14 recommendations in connection with this investigation, although just seven apply to railroads and the railroads’ regulator.

First, I’d like to comment on a dissenting voice, namely that of NTSB Vice Chair T. Bella Dinh-Zarr. Vice-Chair Dinh-Zarr argued that the delay in installing PTC was so gross a failure that the delay can no longer be considered a contributing factor but must be identified as a primary cause for this, and by implication, any subsequent accidents attributable to human error.

This is in a nutshell all that is wrong with the assignment of contributing causes, contributing factors, “likely reasons for,” etc. etc. When we investigate a derailment, or an accident, we are examining the movement of a train. We want to know why the movement was not initiated or completed as intended. And that means determining a cause that is accounted for by the laws of motion. That means identifying what factors, if repeated, when repeated, will always cause a train to derail.

So, not every train lacking PTC will derail at Frankford Jct. We know this because thousands of trains lacking PTC have operated through Frankford Jct. without derailing because the trains operated at or near the authorized speed. The lack or presence of PTC cannot be the cause. We cannot repeat or predict the derailment based on the absence of PTC.

On the other hand, every train operating at or near 106 mph will, or will be at risk to, derail in this curve. We can repeat, predict, calculate the likelihood of derailment based on the speed of operation—that the speed of operation is the determining factor, the cause.

Why do we make this distinction? Because we have locomotive engineers who must control the speed of the train; because PTC may, at any moment, become inoperative; because no matter how more sophisticated the algorithms of enforcement are, they are approximations, estimates, essentially reactive, and we train locomotive engineers to anticipate. Okay, done with that.

Now for the 7 recommendations. First, what is remarkable about them is how little the recommendations have to do with the cause of the derailment. Really? Yeah, really.

NTSB’s first recommendation is to FRA, recommending that FRA require railroads to install in locomotives displays of train location, curves, alignment, elevation of the route to be traveled where PTC will not be implemented. What this recommendation has to do with this derailment or any other derailment of a passenger train is unclear, as all passenger service requires PTC protection.

Moreover, FRA, in its NPRM on minimum crew size, has identified the information contained in such visual displays as a factor that might have a negative impact on the locomotive engineer’s ability to properly execute his/her tasks, a variation of the old “information-overload” argument. So where PTC is not in service, NTSB recommends installation of such displays, while FRA, where PTC is in service, thinks such displays might be a distraction.

This is where speculation, as both NTSB and FRA are speculating regarding the impact of these displays, gets you. George Clinton, the master of funk, knew that a dog chasing its tail will always be busy, going nowhere (see “Atomic Dog,” 1981). I believe any and all railroad managers can be forgiven for throwing up their hands and ignoring both FRA’s concern and NTSB’s recommendation.

The next recommendations advise FRA to alter its accident/incident form to identify the number of crew members in the operating cab or controlling end of a train at the time of an accident.

There is, of course, no evidence that single-locomotive-engineer operation of passenger trains has anything to do with this accident, or the overspeed derailment of 8808 on Metro-North. Moreover, to put it bluntly, crudely and painfully, the history, and right-of-way, of U.S. railroads is littered with the wrecks of trains where two or more crewmembers were present in the operating cab. But . . .

But here’s the thing: NTSB doesn’t need FRA to change its form. NTSB has its own database of railroad accidents, and in every one of the accident reports NTSB has issued that I have read, the report has clearly identified the number of crew members in the operating cabs of the involved trains. So what’s the issue?

Two additional recommendations involve studying and implementing methods to reduce passenger injuries during derailments and overturns, and not even I can argue with that.

But I can argue with the last two recommendations, the “train to multitask” recommendations. Those recommendations have no basis for support in this investigation or in any other NTSB investigation of any other railroad accident. That we can move from speculation to recommendation without accumulated data means the investigation process itself has derailed.

David Schanoes

June 9, 2016