Sunday, July 22, 2012

Engine reliability of Amtrak routes

Numbers are taken from Amtrak's monthly reports, from March 2011 to February 2012. With the exception of the Piedmont, which added a second frequency in 2010 and thus doubled its miles, train miles are as scheduled and calculated in this previous post. So there are some slight inaccuracies, but it shouldn't affect things too greatly. I'm unaware of whether trains annulled due to engine failures are included in Amtrak's reported totals and, if so, how they are recorded.

On a happy note, Amtrak updated its system so that PDFs will actually load in the browser on their site instead of automatically downloading, removing a major frustration of mine. I would not be terribly surprised if they broke all existing external links in doing that however.

I've highlighted the four best and worst performing routes in green and yellow respectively. The Hoosier State is unsurprisingly the worst with its rather low mileage: It would need an almost unblemished record simply to be a mediocre performer.

There does appear to be a severe problem with Amtrak's mechanical department in California however. Discounting the Hoosier State, the three worst performers are based in California and the Sunset Limited is also a Los Angeles based train with significantly worse than average performance (however, the Southwest Chief is Los Angeles based but does not have any outstanding issues). The poor performance of the Surfliner stands in stark contrast to the performance of the state owned Capitol Corridor and San Joaquin fleets, both of which are below the average for engine failure related delays. With Amtrak's locomotives used for the Surfliner averaging 3.5 million miles each (page 8), it's no surprise that there's a mechanical reliability issue.

The performance of the Downeaster, Heartland Flyer, and especially the Missouri River Runner is absolutely outstanding and deserves further investigation. It is imperative that Amtrak form tiger teams with the mission of finding everything that those mechanical crews are doing right, investigating any ways of improving upon these practices, and teaching the rest of Amtrak's mechanical crews those best practices.




Total
Train miles
Minutes per 10K Train miles
Acela
6,747
3,296,176
20.47
NERegional
31,361
5,394,480
58.14




Vermonter
1,895
444,808
42.60
Downeaster
789
422,240
18.69
Keystone
6,417
1,642,680
39.06
Empire
6,488
1,350,544
48.04
Hiawatha
1,192
429,312
27.77
Illinois Services
4,948
1,654,016
29.92
Michigan Services
5,292
1,024,296
51.66
Missouri Services
195
412,048
4.73
Heartland Flyer
189
149,968
12.60
Pacific Surfliner
11,371
1,632,800
69.64
Cascades
4,268
1,086,176
39.29
Capitols
3,397
1,201,928
28.26
San Joaquins
4,862
1,326,416
36.66
Hoosier State
729
81,536
89.41
Carolinian
1,601
512,512
31.24
Pennsylvanian
1,295
323,232
40.06
Piedmont
1,081
251,888
42.92




Silver Star
4,297
1,011,192
42.49
Cardinal
1,649
357,864
46.08
Silver Meteor
4,723
1,011,192
46.71
Empire Builder
6,483
1,897,168
34.17
Capitol Ltd.
2,195
567,840
38.66
California Zephyr
12,990
1,774,864
73.19
Southwest Chief
8,143
1,642,368
49.58
City of New Orleans
1,685
674,128
25.00
Texas Eagle
4,209
950,040
44.30
Sunset Ltd.
3,453
622,440
55.48
Coast Starlight
8,618
1,002,456
85.97
Lake Shore Ltd.
4,268
843,024
50.63
Palmetto
1,552
603,512
25.72
Crescent
4,491
1,002,456
44.80
Auto Train
1,857
622,440
29.83




Corridor Average
87,370
19,340,880
45.17
LD Average
70,613
14,582,984
48.42

4 comments:

  1. This seems really strange to me. As far as I can tell, there's a lot of deviation between routes which are generally served by the same shops, but unfortunately I'm not familiar enough with Amtrak's maintenance facilities to know for sure -- shouldn't routes out of Chicago see roughly similar numbers, for instance?

    The Downeaster, Missouri River Runner, and Heartland Flyer would all seem to have shops independent of the main network. They each only have 1 or 2 trains worth of equipment to maintain, right?

    I used to work as a software QA tester sending network traffic through firewalls -- I generally only accepted tests if they had success rates of 99.995 to 99.999% or better (and sometimes, even a one-in-a-million issue was something that should get corrected). Despite some similarities between railroad lines and computer networks, I'm not sure how that would translate to rail -- my gut tells me that we really should be seeing engines have only 1 or 2 minutes of downtime per 10k train miles.

    That might be a bit ambitious, but it certainly seems like an order-of-magnitude improvement for most services (getting them under 10 minutes each) would be feasible and very beneficial.

    ReplyDelete
    Replies
    1. I'm not sure why some routes have greatly different numbers despite the same shops. Three thoughts come to mind: 1) Some trains always run with particular engines, even if they could be switched out, and so they'll be more prone to breakdowns than other trains 2) Location and availability of protect locomotives may help, though I think this would only be a minor factor. 3) Certain routes may run with two locomotives as standard (for lack of a suitable wye or cab cars) and so they may be able to keep running through the failure of one (although I'm not sure that that's the case, some of the freights have required Amtrak to run with at least two though on their territory due to engine reliability).

      Unfortunately, I don't have data for commuters or rest of the world systems, but I think your gut is probably in the right area. In one old report on checking oil more often for various organizations, Canadian Pacific is mentioned as having "40 or so engine failures per year" back in the 1980s. Now, admittedly, Amtrak's locomotives probably put more miles on them and have a more stressful use cycle, but I'd be willing to bet that the Surfliner alone has more engine failures than that.

      Delete
  2. This doesn't look like maintenance practices to me.

    This looks like operating practices. The trains doing the best have a captive fleet. The trains doing the worst are at the same maintenance bases as trains which do just fine.

    Amtrak has a shortage of engines (yes, really) and they're all old (for diesel engines, which sadly wear out fast). This looks to me as if what happens is that certain trains get the "best engine in the yard" and certain others get "whatever's left over".

    --Nathanael

    ReplyDelete
  3. Just to dogpile on this long after publication, at least one of those routes has issues unrelated to maintenance; the Zephyr runs through a portion of Iowa that got an ARRA grant to install crossovers to reduce delays, which means there's a congestion issue there.

    I remember reading about the Ethen Allen having an abysmal on-time record, which was resolved (iirc, a 1000% reduction in minutes of delay) by fixing outstanding track issues.

    Not to say that engine maintenance issues aren't there, but that plucking a single statistic with multiple causes isn't an apples-to-apples comparison.

    ReplyDelete