ON-TIME PERFORMANCE: 5/13/04 THROUGH 6/11/04

The results for 30 dates in the most recent period are:

TOTAL TRAINS = 1020

ON TIME = 183 (18%)

LATE = 837 (82%)

AVERAGE DELAY = 1' 10"

Only scheduled arrival times were used. No "grace" period was allowed.

Of the late trains:
279 (33%) were 29" or less late;
184 (22%) were 30" to 59" late;
219 (26%) were 1' to 2' late;
141 (17%) were more than 2' late; and
14 (2%) had no arrival times shown.

BY ROUTE, WITH O.T.P. AND AVG. DELAY:

CAPITOL LIMITED
#29 - 10% - 2' 12" #30 - 0% - 1' 47"

THREE RIVERS
#41 - 30% - 1' 30" #40 - 7% - 1' 09"

PENNSYLVANIAN
(W) - 20% - 51" (E) - 3% - 37"

LAKE SHORE LIMITED
#49 - 0% -2' 12" #48 - 7% - 2' 04"

LAKE SHORE LIMITED (BOSTON / ALBANY)
#449 - 0% - 1' 04" #448 - 0% - 1' 53"

CITY OF NEW ORLEANS
#59 - 33% - 1' 29" #58 - 30% - 49"

CRESCENT
#19 - 0% - 54" #20 - 57% - 52"

DETROIT CORRIDOR
(W) - 26% - 49" (E) - 31% - 49"

ST. LOUIS CORRIDOR
(W) - 16% - 51" (E) - 6% - 1' 29"

SYRACUSE CORRIDOR
(W) - 28% - 41" (E) - 17% - 1' 07"


The results of this period are most disappointing with the sharp decrease in OTP and the increase of average delay to more than 1' versus the prior periods.

The percentage of trains less than 1' late declined by a third, while that of the trains more than 2' late doubled.

Of the trains more than 2' late, the Lake Shore Limited accounted for 40, the eastward St. Louis Corridor for 18 (14 of which were #22 Texas Eagle), the Syracuse Corridor for 18, the Capitol Limited for 15, the Three Rivers for 10, the City of New Orleans for 9, the Detroit Corridor for 8, and the Crescent for 7. The increase in the St. Louis Corridor is due to not operating an on-time section of #22 from St. Louis on the days when #22 was operating very late and to the late operation of #304 from KCY to STL.

The operation of the westbound Lake Shore Limited was disrupted by the failure of the Boston section (#449) to achieve a single on-time arrival over CSX to Albany . The operation of the Lake Shore Limited combined sections and of the Capitol Limited in both directions between Cleveland and Chicago continued to lose time on more than half the trips over this Norfolk Southern route.

The eastbound Pennsylvanian and 2 trains on the St. Louis Corridor (#21 and #304) managed to post only a single on-time arrival, while the eastbound Lake Shore Limited, the eastbound Three Rivers, and #49 and #64 on the Syracuse Corridor were on-time only twice each. Given the 1' 30"-plus padding in the schedule of the Three Rivers, this failure is surprising.

At least 65 trains departed their initial terminals (BOS, NYP, PNT, CHI, NOL, KCY) so late that an on-time arrival was not possible. This number is also an increase over prior periods. 28 of these departures were more than 1' late, 9 of which more than 2' behind schedule.

Even allowing for delays due to equipment shortages, severe weather related problems, and freight derailments, this was not at all a good period.

eg,

Your query isn't silly at all. The figures referred to in the other (VerMontanan's) post about OTP are presumably Amtrak's own. As such, they include the allowance of the grace period, and this ranges from 15" to 2', depending upon the runs involved. As stated in the opening of this review, I use the timetable arrival times and do not allow for any grace period. Late is late, period. Also, the other report is for a calendar month which is not the same period as in this review.

As to the figures which you mentioned, they are from a specific host railroad and cover only the portion of the run or service which is over that particular host railroad. For example, the Lake Shore Limited operates over Amtrak territory between Penn Stn. (NYP) and Spuyten Duyvil; Metro North between there and CP-75 north of Poughkeepsie; CSX between CP-75 and DB in Cleveland; and Norfolk Southern from Cleveland to Chicago. If NS makes the scheduled running time over its portion of the run, it receives credit for an on-time train regardless if the train is not on its schedule. Thus, if Amtrak holds #48's departure for 45" and NS delivers the train to Cleveland 45" late, that is an on-time run for NS.

The issue of "padding" in the schedule (aka as Amtrak "recovery time") is different and is not involved in figuring the on-time performance. The padding is simply part of the schedule. Some of the long-distance trains will have padding at more than one place in the schedule in addition to the approach to the final terminal. #30 eastbound "Capitol Limited" has about 30" padding between Cleveland and Pittsburgh and another 30" between Rockville and WAS.

Hope this clears thing up.

These results are absolutely abismal. Very aware of the complicated relationships that Amtrak has to bear with its host railroads, but someone at Amtrak needs to confront these poor statistics - as ontime trains to continue to be a cr*p shoot. It's not healthy and more than likely producing a lot of "never again" riders, which is a downward spiral. The Lake Shore Limited ontime performance increased tremendously a couple of months ago (when the Boston section was operating as a stub end train). Now the Lake Shore is back to its old ways.

If Amtrak could get rid of the padding, convince its host railroads to operate trains on time (simpler than it sounds) and get some stability in the schedule, train travel will become somewhat more attractive. Instead - hit or miss onboard service and hit or miss timekeeping. Not good business.

These one time performances are an absolute disgrace. would seem to me
that some sanctions against the worst freight railroads should be assessed
by the "feds". Any funds collected to go directly to Amtrak.
Years ago, the railroads fought to get out of the passenger business, so
finally Amtrak was created to relieve them not of the trains but of the
deficits involved in their operation.
The fact that the freight railroads have cut back on capacity and that
freight business is up should not rest on the backs of Amtrak's passengers.
It seems to be the same freight railroads that are the worst offenders
month after month.
Noel Weaver

After looking at the previous posts, it seems to me that the best on time
performance for a single long haul freight railroad has to be BNSF. The
BNSF runs loads of freight trains over lines that have a lot of single track
operation and seems to be able to do better with both their freight trains
and with Amtrak.
Canadian Pacific is another railroad with a lot of single track but they seem to have their act together.
I have seen a lot of Guilford bashing on these forums but they seem to do
better than average too. I know I rode the "Downeaster" last summer and we arrived Portland on time and the only reason we were late on the
return to Boston was a massive electrical storm which knocked out a lot of
signals. I have seen this happen on many different railroads so I can't in
good conscience pin this delay on Guilford. While I was riding the DE,
Guilford held a freight train in both directions for us to go first, I wonder
what UP or CSX would have done in the same circumstances.
Noel Weaver

If Amtrak could get rid of the padding, convince its host railroads to operate trains on time (simpler than it sounds) and get some stability in the schedule

Congress has a role to play here. Getting Amtrak off the annual appropriations merry-go-round is an important element. Until Congress sends a clear message that Amtrak is here to stay, then the host railroads will have little external incentive to handle its trains properly. And in fact just the opposite: Late trains drive passengers away, and lightly patronized trains get taken off. I'm not saying every host's managers act on that temptation, but the temptation does exist as long as Congress denies Amtrak a long term. Write.

As for stability in the schedule, that's happening. Train schedules in recent years haven't displayed much, but there was far less disruption at the spring/sunmmer change than at the previous half dozen.

After looking at the previous posts, it seems to me that the best on time performance for a single long haul freight railroad has to be BNSF.

That's one reason I call this summary good for bashing and not much else. It's not the just no-mercy rule, it's the small and skewed sample: mainly eastern trains, endpoints only. There are lots of trains missing. For example, no BNSF-hosted trains appear on it.

More comprehensive and thus more useful reports on on-time performance now appear on Amtrak's website, here

I agree with Mr. Weaver about the feds' having more of a role in OTP enforcement. I say that's well within the scope of the Federal Railroad Administration. Writing the new administrator, when one is seated, would be worth doing. FRA should be more like FAA; the right administrator can nudge it in that direction.

Concur.

Incomplete data is worse than no data.

Mr Lustig, please post system wide data, including samples of intra-trip timeliness (such as Denver to SLC or Saint Louis to New Orleans).

John_Perkowski wrote:Mr Lustig, please post system wide data, including samples of intra-trip timeliness (such as Denver to SLC or Saint Louis to New Orleans).

Or, you could do it yourself. I get the impression that he gets all his data from Julie/Amtrak's website, and compiles it himself. That's a heckuvalotta work. A couple of years ago, I was planning a trip, and tracked the OTP of several trains I was going to be riding. Just that small sample (10 trains) took quite a bit of time, and I was only checking the stations that applied to me.

You want details for more trains, and for intermediate stops?
http://tickets.amtrak.com/Amtrak/trains ... lCall=true

There you go.

I get the impression that he gets all his data from Julie/Amtrak's website, and compiles it himself. That's a heckuvalotta work.

True enough, and thankfully no longer necessary in its current form. A constructive way to redirect those efforts would be as Mr. Perkowski describes: OTP not only at endpoints but at key midpoints as well. But to keep the task manageable, work with a smaller sample of selected trains. Here's a way to compile 900 data points in a month's time:

1. Train 48: South Bend, Toledo, Cleveland, Buffalo, Rochester, Rensselaer, New York (7 x 30 = 210)
2. Train 49: Rensselaer, Rochester, Buffalo, Cleveland, Toledo, South Bend, Chicago (7 x 30 = 210)
3. Train 30: South Bend, Cleveland, Pittsburgh, Washington (4 x 30 = 120)
4. Train 29: Pittsburgh, Cleveland, South Bend, Chicago (4 x 30 = 120)
5. Train 59: Champaign, Memphis, Jackson, New Orleans (4 x 30 = 120)
6. Train 58: Jackson, Memphis, Champaign, Chicago (4 x 30 = 120)

That's less work (900 v. 1,020), and 80% of it is unduplicated. Looks like a winner to me.

Howdy Amtrakers,
Speaking of on time performance.I was waiting on 2:00 regional from
Newark to Boston yesterday and it was over an 1 1/2 hpurs late getting in.
In fact the 3:00pm came in just before it so I got on that not knowing
86 was right behind it.Conductor had 0 info.While waiting up on Track 2
it was announced that all Eastbound service had been suspended due to police activity,but what that had been was never specified.Talk among
folks on platform naturally considered possible terrorist related incident
but nothing in around Newark station suggested that at all.Anyone
have any idea what it was all about?Gotta give crews credit though,the local I ended up getting made it to Boston from Newark in about 4 hours
and 20 minutes with stops everywhere but Mystic.Anyway would appreciate any info,as I had so wisely upgraded to reserved since I was able to get to Newark in time for 86.LOL!!!

Ontime performance at intermedaite points is likely to be worse , than ontime performance at endpoints . thats because there is padding before the endpoint on all train schedules .

What's great about putting train status online is that anyone can check it. Anywhere. Even half a world away. Howazabout testing that hypothesis with a month's worth of data?

That would be interesting , however i dont have the time to do that . But i think its logical to assume for example , that the Sunset having 56 minutes of padding between its last stop (Pomona) , and L.A ,if it was late at its endpoint of L.A , it was even later at Pomona . Given that it is allowed 3 hours and forty minutes to traverse its last 106 miles , ( 29 mph average ) versus 2 hours , 24 minutes outbound ( 41 mph average ).
Ive never understood the logic of padding the schedule . Perhaps it would be better for Amtrak to remove the padding , and let the ontime statistics show the full impact of railroad delay .
i actually got got out one or 2 times by the padding . once on the empire builder bound for Seattle , i awoke to find us around 3 to 4 hours behind schedule . this would have put us arriving in Seattle around 2 in the afternoon , so i settled in to spend most of the day on the train . to mt surpirse we were in Seattle about 1 hour later .
The other time i was awaitnig the arrival of the train in Truckee , actually rang to find out when the train was due , was running around an hour late , so i took my time getting my bicycle ready for carrying on the train , and lo and behold it arrives ahead of time . i did make it though seem to remember leaving 1/2 my breakfast behind .

David Benton wrote:That would be interesting, however i dont [sic] have the time to do that .

Don't have the time? Let's not kid ourselves: Anyone who has the time to post here an average of 1.16 times a day can easily make the time to test his own hypotheses. Anyone who can master the task of posting here can easily use amtrak.com to check train status.

Three trains is too much? Fine. Pick one.

David Benton wrote:Ive never understood the logic of padding the schedule . Perhaps it would be better for Amtrak to remove the padding , and let the ontime statistics show the full impact of railroad delay.

Well, then what's the point of having a schedule at all? Why not, "We get there when we get there."

They pad the schedules at major service stops and at the end of the route because one expects to have some delay en route. It's just really hard to predict where that delay will occur (and indeed, the points where a train gets held up are often different from one day to the next). Therefore, by padding the endpoint times a bit, you can accommodate the train's lateness (to a certain extent) regardless of where that delay happened.

For example, my southbound Coast Starlight (admittedly not known for its on-time performance anyway) a month ago was an hour or so down at Santa Barbara (I don't know the exact times since I can't find my trip notes at the moment). However, we only arrived into Los Angeles 10 or so minutes late. That train lost some time in Oakland for a medical emergency, and a little later on due to opposing rail traffic. On other days, that train might have lost a bit of time south of there due to traffic, or maybe north of Oakland, etc.

The point is, delays are expected en route, and the padding accommodates this. Why? Because (unlike dedicated Amtrak devotees) most people interpret a schedule as the time they will actually be getting somewhere. As much as we'd like to, it's not practical for Amtrak to have every intermediate time point accurate 100% of the time. Most of them represent the best-case scenario for running between two intermediate stations. However, Amtrak realizes that the "best-case scenario" doesn't happen all the time, and something's going to get in the way. Therefore, by adding padding, Amtrak can offer passengers arriving at the train's last stop a better chance of arriving "on time" regardless of whether the train was delayed leaving its initial terminal, or if it happened to get delayed 5 miles from the final destination.

Why else should Amtrak pad its schedules? Connections. If someone wanted to take train A, which arrives at XYZ at 3:30 (without padding), and connect to train B, which leaves at 4:30, then every time train A arrived an hour late (assuming a guaranteed connection), Amtrak would have to pay to reaccommodate that passenger. However, if the padded schedule has train A arriving at 4:15 instead, then the connection is no longer valid, so if train A arrived a bit late, then they wouldn't be responsible for the passenger who misconnected. Of course, this is a very simplistic example, but it just shows that schedule padding is useful for allowing trains a better chance to arrive on time, and a better chance of making the "guaranteed" connections (otherwise, Amtrak would have to up the time for a guaranteed connection from a LD train to around 2 hours, from the current 1 hour...or is it 90 minutes?).

Even with schedule padding, though, some long-distance trains struggle to make their end points on time.