OTP and Delay Accountability

[NSC1109]

We all know that Amtrak's OTP has really gone down the crapper in recent months. Take a gander at the sample statistics below, copied from our sister forum railroad.net and originally created by user "shlustig", who posts these every month for select corridors around the network:

https://railroad.net/forums/viewtopic.php?f=46&t=169890

SELECTED ON-TIME PERFORMANCE 6/1 THROUGH 6/30

Post by shlustig » Tue Jul 02, 2019 1:47 pm

Results for this period are:

TOTAL TRAINS = 1044

ON TIME = 307 (29%)

LATE = 737 (71%)

AVERAGE DELAY = 58"

Of the late trains:
40% were 29" or less late;
25% were 30" to 59" late;
22% were 1' to 2' late;
12% were more than 2' late; and
>1% had no arrival time shown.

BY ROUTE, WITH ON-TIME % AND AVERAGE DELAY:

CAPITOL LIMITED
#29 - 3% - 1' 51"
#30 - 3% - 2' 02"

CARDINAL (WAS)
#51 - 23% - 47"
#50 - 23% - 1' 03"

PENNSYLVANIAN
#43 - 17% - 39"
#42 - 50% - 50"

LAKE SHORE LIMITED
#49 - 21% - 1' 40"
#48 - 28% - 1' 09"

LAKE SHORE LIMITED (BOSTON / ALBANY)
#449 - 57% - 1' 03'"
#448 - 25% - 2' 52"

CITY OF NEW ORLEANS
#59 - N/A
#58 - N/A

CRESCENT
#19 - 3% - 1' 21"
#20 - 20% - 1' 21"

DETROIT CORRIDOR
(W) -4% - 41"
(E) - 17% - 37"

ST. LOUIS CORRIDOR
(W) - 53% - 31"
(E) - 49% - 1' 12"

BUFFALO CORRIDOR
(W) - 7% - 43"
(E) - 49% - 47"

KANSAS CITY CORRIDOR
(W) - 30% - 1' 43"
(E) - 25% - 55"


Only 85% of the full complement of trains was reviewed in this period due to cancellations caused by flooding, programmed trackwork, and other operational incidents primarily on the routes of the City of New Orleans, Boston section of the Lake Shore Limited, and KCY Corridor.

Services which showed improvement in both measured categories included the westbound Pennsylvanian and STL Corridor, northbound crescent, and eastbound DET Corridor.

Of the 86 trains that were more than 2' late: the Capitol Limited accounted for 23, eastbound Texas Eagle on the STL Corridor for 15, Lake Shore Limited for 12, and the Crescent for 11.

Of the total delay minutes in this period: the STL and BUF Corridors each accounted for 18%, the DET Corridor and the Capitol Limited for 15% each, the Lake Shore Limited for 12%, and the Crescent for 10%.

The dismal level of substandard performance continues with no relief in sight.

Clearly, OTP is at a substandard level. What is not so clear (at least not all the time) is why it is at the levels we see here.

First off, Amtrak needs to get a handle on delays. We all know that most of the delays occur while on freight-owned tracks. We also know that certain companies are better than others when it comes to OTP. Amtrak needs to analyze the following:

1) What are the root causes of delays for this service/line?

2) Are those delays easily avoidable? What operations are interfering with our OTP? Can we change our operations to mitigate or eliminate delays?

3) How can we work with our host company to mitigate/eliminate delays? (or, in the case of the AML, what can we do to mitigate/eliminate delays?)

4) How can we be sure that we are informing not only our passengers but their friends and family about delays accurately and clearly?

The answers:
1) Root Cause Analysis is something I learned in my Six-Sigma Problem Solving course I finished a few weeks ago. The idea is to keep asking "why" until you end up going in circles. (That is a very basic way of putting it, just FYI).

For example:
365 was late arriving in BTL.
WHY? --> CN routed a landbarge ahead of us.
WHY? --> We were late departing PTH, so we missed our slot.
WHY? --> The boarding process took too long.
WHY? --> There were more people than normal.

The red line indicates the root cause of 365's delayed arrival into BTL. Now, Amtrak can take that root cause and make changes in the boarding process to ensure that 365 is able to make its time slot.

2) Delays that are easily avoidable are the worst kind of delays. That being said, Amtrak should implement (if they haven't done so already) a comprehensive delay-coding-and-tracking system to monitor the cause of every single delay. These delays can be analyzed by NOC in Delaware where common occurrences can be identified and RCA can be implemented to find the root cause of the delay and changes can be made to avoid those issues. We use a similar system at Delta Airlines, where we have a code for literally every single type of delay plus a text box to give specifics. For example, one flight we had was given delay code 03A - A/C MECHANICAL with text box description GPS FAIL TO ALIGN causing a 30-45 minute delay in the flight's departure. This type of delay tracking can help across all departments, providing customers with accurate information regarding delays, mechanical teams being notified of an issue before the train (or aircraft) comes in, etc. It also helps with accountability, ensuring that a delay can be traced back to its source and corrective action taken.

3) Amtrak pays a "token payment" to host railroads for trackage rights and as a result, the hosts are not exactly inclined to provide superior service. To that end, I propose to end the "host" moniker and instead turn it into a partnership so that when Amtrak is on time, all parties win. This will undoubtedly mean higher payments to freight companies, but I believe that the potential for increased OTP would be worth it. One of the chief complaints I see from the public as to why they don't like taking Amtrak is that you never know if you're going to arrive on-time. Once you are able to turn that reputation around and bring OTP consistently into the 80-90% range, you will likely see more riders. More riders = more revenue = potentially less subsidy (for the "fiscally conservative" types out there). But it wont happen unless the investment is made. It's like I tell people who complain about the bad quality of the roads...you want it to be fixed but you don't want to pay for it (via taxes). Then it's not going to get fixed.

4) Again, the delay-tracking system would work wonders here. Right now, if you go to the Amtrak twitter page, you'll see a jumble of various reasons why trains are late that are seemingly similar to each other, but the difference is apparently known only to Amtrak. For example, I frequently see "freight railroad congestion" and "railroad congestion" both used when posting about delays. Not only that, but sometimes the posts are almost downright incorrect. One recent post (a few weeks ago) regarding 354 talked about an "unforeseen crew change in Centrailia", which isn't anywhere close to 354's route. The delay-tracking system could ensure that customers and their family/friends are receiving accurate information regarding issues during travel.

If you've made it this far, I sincerely thank you for taking the time to read this. I know it's a lot of information, but my time thus far with Delta as a Ramp/Operations Agent (and too much time on my hands during the offseason for school) has helped me come up with some ideas that could potentially improve Amtrak service.

For those that know far more about internal operations than myself, a lowly college student, please feel free to add or correct anything that is already used.

-NSC1109

 

[RebelRider]

Amtrak does have a comprehensive delay tracking and coding system in place. Conductors have an application called eDR (electronic Delay Reporting - clever, right?) on their mobile devices. We are required to report a reason for every delay that exceeds one minute. If the delay is one minute and caused by rounding, nothing has to be reported. If a one minute delay did occur then it must be reported.

As far as coding, we have lots and lots of codes along with free form comments. A 40 page manual describes pretty much every possible scenario so conductors can correctly code these delays.

There are 27 top-level and 167 second level delay codes. When coding the “most-direct cause” of a delay should be reported. We can also report multiple delay codes for any reporting segment. For example, a train is delayed 24 minutes between point A and B. Two minutes can be attributed to a temporary speed restriction (DSR), 8 minutes on weather (WTR) due to heat orders and 14 minutes following a freight train (FTI).

At a minimum, the actual location the delay occurred must be included. It can be a single milepost location, a range between certain mileposts, designated control points, siding names, etc. At locations where host railroads change we must report which railroad delayed the train. When meeting other trains the symbol or engine number is required. Speed restrictions require the location and speed. Signal delays must report the signal aspect received that caused a delay. Others trains can be reported as Amtrak (PTI), commuter (CTI) or freight (FTI). Mechanical delays must include the car or engine number causing the delay along with a description of the issue. If we bounce back and forth between tracks or in and out of sidings with no apparent reason, we code the delay as routing (RTE). Maintenance of Way (DMW) could involve waiting for workers to clear off the tracks or an inability to reach the MOW foreman for permission to proceed.

My personal favorite is NOD which stands for No Delay. This is typically used at stations where recovery time allowed for an early arrival and the train exceeded the scheduled dwell minutes waiting to depart on time.

I could go on. As conductors enter delays, they become immediately viewable by anyone with access to the eDR data. How that information is being crunched later to analyze and mitigate delays is outside my realm. But it is being captured and is what drives the host railroad OTP reports.

 

[NSC1109]

Amtrak does have a comprehensive delay tracking and coding system in place. Conductors have an application called eDR (electronic Delay Reporting - clever, right?) on their mobile devices. We are required to report a reason for every delay that exceeds one minute. If the delay is one minute and caused by rounding, nothing has to be reported. If a one minute delay did occur then it must be reported.

As far as coding, we have lots and lots of codes along with free form comments. A 40 page manual describes pretty much every possible scenario so conductors can correctly code these delays.

There are 27 top-level and 167 second level delay codes. When coding the “most-direct cause” of a delay should be reported. We can also report multiple delay codes for any reporting segment. For example, a train is delayed 24 minutes between point A and B. Two minutes can be attributed to a temporary speed restriction (DSR), 8 minutes on weather (WTR) due to heat orders and 14 minutes following a freight train (FTI).

At a minimum, the actual location the delay occurred must be included. It can be a single milepost location, a range between certain mileposts, designated control points, siding names, etc. At locations where host railroads change we must report which railroad delayed the train. When meeting other trains the symbol or engine number is required. Speed restrictions require the location and speed. Signal delays must report the signal aspect received that caused a delay. Others trains can be reported as Amtrak (PTI), commuter (CTI) or freight (FTI). Mechanical delays must include the car or engine number causing the delay along with a description of the issue. If we bounce back and forth between tracks or in and out of sidings with no apparent reason, we code the delay as routing (RTE). Maintenance of Way (DMW) could involve waiting for workers to clear off the tracks or an inability to reach the MOW foreman for permission to proceed.

My personal favorite is NOD which stands for No Delay. This is typically used at stations where recovery time allowed for an early arrival and the train exceeded the scheduled dwell minutes waiting to depart on time.

I could go on. As conductors enter delays, they become immediately viewable by anyone with access to the eDR data. How that information is being crunched later to analyze and mitigate delays is outside my realm. But it is being captured and is what drives the host railroad OTP reports.

I suspected that Amtrak had something in place given the posts on Amtrak Alerts, bud I wasn’t sure how “comprehensive” it really was. I certainly hope that someone, somewhere is using the data to get to the bottom of things.

Thanks for providing an “insider view”!

 

[Old MM]

I know ALLLLL about delay coding in the airline world, at a very granular level. God I hate the delay coding system.
The reason for its existence is to get your personnel to bust a gut meeting departure times. If equipment is late, place additional stress on personnel to turn the equipment quickly. This can, and does, result in service deficiencies. More in the airline world than for trains I am guessing.
Of course, if you didn't pressure your people, the equipment would not turn quickly, but would continue to incur additional delay.
Weather and mechanical issues cause most delays. Trying to blame the people who are trying to do their jobs is the brainchild of people in charge.

 

[Trogdor]

The challenge with all of this is: so what?

The fact that freight train interference is a major cause of delay on certain routes has been known since forever, has been made public, the railroads responsible have been identified and shamed, and...still, they don’t care. Those very same freight railroads also have to approve every single schedule change affecting a train on their territory. Amtrak can’t unilaterally implement a new schedule if it changes the time a train would operate over a host railroad’s tracks. When Amtrak proposes changes, host railroads may wait months (or longer) before they approve, if they approve at all. Any change that even hints at taking time away (without hundreds of millions in additional money and a specific service outcomes agreement with the host stating that time will be removed from the schedule) will almost automatically be rejected. Adding time to a schedule doesn’t really work either (look to the Canadian for an example of a route that has had a ton of time added, and still runs ridiculously late).

The fact that a passenger train is scheduled through the same place at the same time every day is not a surprise to the host railroad, yet they will still delay it.

Legal recourse against the host may be an option (and has been tried at times in the past), but beyond that, there’s not a lot that can really be done from Amtrak’s side, since so much of it is beyond their control.

 

[NSC1109]

The challenge with all of this is: so what?

The fact that freight train interference is a major cause of delay on certain routes has been known since forever, has been made public, the railroads responsible have been identified and shamed, and...still, they don’t care. Those very same freight railroads also have to approve every single schedule change affecting a train on their territory. Amtrak can’t unilaterally implement a new schedule if it changes the time a train would operate over a host railroad’s tracks. When Amtrak proposes changes, host railroads may wait months (or longer) before they approve, if they approve at all. Any change that even hints at taking time away (without hundreds of millions in additional money and a specific service outcomes agreement with the host stating that time will be removed from the schedule) will almost automatically be rejected. Adding time to a schedule doesn’t really work either (look to the Canadian for an example of a route that has had a ton of time added, and still runs ridiculously late).

The fact that a passenger train is scheduled through the same place at the same time every day is not a surprise to the host railroad, yet they will still delay it.

Legal recourse against the host may be an option (and has been tried at times in the past), but beyond that, there’s not a lot that can really be done from Amtrak’s side, since so much of it is beyond their control.

That’s why part of my post talked about making it worth their while to make trains run on time. Without knowing more of the agreements between Amtrak and freight companies, I can’t speculate too much, but I would think that some sort of “partnership” that would benefit both parties if trains ran on time say 90% or higher. There would be a lot of work in it but I believe it would pay off in the end through potential increases in ridership and eventually revenue.

 

[Trogdor]

That’s why part of my post talked about making it worth their while to make trains run on time. Without knowing more of the agreements between Amtrak and freight companies, I can’t speculate too much, but I would think that some sort of “partnership” that would benefit both parties if trains ran on time say 90% or higher. There would be a lot of work in it but I believe it would pay off in the end through potential increases in ridership and eventually revenue.

The problem is that there is virtually no scenario that would make running reliable passenger service worthwhile to a freight railroad. Some time back, I dug up some numbers showing the discrepancy between the revenue of a freight train vs. a passenger train.

Could Amtrak Be Replaced With Something Better?

This was in the context of having freight railroads take over the operation of passenger trains, but the general concept applies here as well. I don’t know the numbers offhand, but the money Amtrak pays host railroads is a tiny fraction of what they could generally earn with freight trains. In order for a passenger train to be beneficial to a freight railroad, it would have to earn at least its opportunity cost to the freight railroad (which is not only the potential profit they could earn from running another freight train instead of a passenger train in that slot, but also make up for the lost capacity that inherently comes with running trains of differing speeds on the same line and giving the passenger train a general priority). I don’t want to say that will never happen, but I will say it’s hard to envision a scenario in which that happens that doesn’t so greatly inflate the cost to Amtrak of running that train that it becomes nearly impossible for any but the most extreme rail supporters to justify the subsidy.

The freight railroads figured out a long time ago that it’s not worth providing more than the bare minimum accommodation to Amtrak, simply because they’re required to by law. The big unknown has always been the extent to which they can (or will) be punished for not getting the trains over the road on time. History has shown the answer to be, not much. So they don’t.

If changing that calculus was easy, it would have been done already.