MARC engine sitting in the Silver Spring Station on the Brunswick Line in 2012 by Michael Wilson used with permission.

The Maryland Transit Administration recently announced plans to spend up to $1.6 billion to replace train infrastructure on its three MARC lines as part of its MARC Cornerstone Plan. However, I believe they aren’t spending their money wisely, with existing plans to buy more trains that will run slowly and on tracks that need to be maintained at a higher standard of service.

Taking the Brunswick Line from Frederick to Washington, DC, is the most unnecessarily slow trip offered by MARC. While this saves the rider a long drive where they are more likely to experience or cause a traffic fatality – the trip is a plodding 130 minutes, compared to 90 minutes driving during rush hour. Trips from Martinsburg, the western end of the line are similarly slow at 127 minutes. While the other two MARC lines are better – the Camden Line takes 75 minutes instead of 70 minutes by car at rush hour from Camden Yards to Union Station, and the Penn Line takes 109 minutes versus 120 minutes by car from Perryville – the Brunswick Line is unique in how slow it runs.

A big reason why the Brunswick Line takes so long is that its trains are a bad fit for its service. Locomotive-hauled trains, particularly those with diesel locomotives, are notorious for sluggish performance. Each stop takes an inordinate amount of time to slow down on approach and accelerate back up to speed, a problem for the densely-packed stations of the Brunswick Line.

The Brunswick Line in particular is a lost opportunity. According to MDOT, 480,000 jobs are within 15 minutes of stations. It connects the two sides of the Red Line; the second- and third-largest cities in Maryland (Frederick and Rockville); and exceptionally walkable, rail-oriented cities.

While the rest of the world has transitioned to the train equivalents of jet planes, with fast acceleration and deceleration and high performance, MARC has kept its clunker trains around, flying the train equivalent of 1930s-era propeller planes.

I believe that the lack of specification of equipment type in the part of the Cornerstone Plan that calls for new trains doesn’t solve this issue. Instead I think it allows MARC to operate how it is used to, buying newly-made clunkers to replace the old ones. This will squander a rare opportunity to improve service without any new funding. No decision has been made, but I would like them to decide differently than they have in the past.

The jets of the train world are multiple units (MU), which are usually either electric multiple units (EMUs) or diesel multiple units (DMUs). Unlike locomotive trains, where the engine sits in front or behind the unpowered carriages, each MU car has its own engine. This lets them accelerate and decelerate faster. While all trains can tilt into curves, MUs pair that with an ability to accelerate quickly out of these slower curves. EMUs, like those that run on Metro, are lighter than DMUs and so perform even better.

A second reason is poor track maintenance. While the geometry of the Brunswick and Camden lines could allow for at least 79 mph service, if not 125 mph, in practice they slow dramatically in places. This is worst between Point of Rocks and Frederick, which slows to just 45 mph on the Old Main Line through most of Frederick County and hardly more than 30 mph on the Frederick Branch itself. With improved maintenance, these could open up much faster service without new track or capital outlay.

Why MARC hasn’t done either of these things has not been stated outright, but there are some hints. Officials may be unwilling to change up train maintenance practices: MARC is familiar with how to maintain and operate locomotive-hauled trains, but not DMUs or EMUs. Retooling and retraining maintenance and operations staff to work with a new kind of train is disruptive and means abandoning years of experience.

Track maintenance is another issue, as CSX, a freight operator, owns almost all of two of the three lines (Camden and Brunswick; Penn runs on Amtrak-owned track). They would need to be convinced to improve track maintenance. The one bit owned by MARC, a small segment between Monocacy and Frederick stations, would also need upgrades. Given that so-called State of Good Repair (“SOGR”) funding can be a black box for the public and oversight bodies, the cost of upgrades, annual costs to maintain upgrades, and the line-item breakdowns of each, should be given upfront.

Modeling service improvements with faster trains

To see how new trains and better tracks could pair up, I built a model of the three MARC lines and checked to see how different DMUs and EMUs might operate on them at 79 mph, 110 mph, and 125 mph.

To do this, I put in my model some key operating characteristics (top speed, average acceleration to top speed, and average deceleration from top speed) of some of the top-performing trains in the world as well as some familiar trains from the United States. These are:

  • British Rail Class 350 Desiro, built by Siemens and operated by West Midlands Trains in the United Kingdom as short-hop regional trains.
  • British Rail Class 395 Javelin, built by Hitachi and operated by Southeastern in the United Kingdom as high-speed commuter trains.
  • The Japanese Shinkansen N700, a Japanese-developed train, and the fastest high-speed Shinkansen train, operated all over the country.
  • The Stadler KISS, a Swiss-developed train that is currently planned to run on Caltrain between San Francisco and San José and already operating all over Europe for regional rail.
  • The Stadler FLIRT, a smaller train than the KISS developed for regional rail use and deployed all over Europe, from Azerbaijan to Portugal
  • Acela Express, which operates higher-speed service along Amtrak’s Northeast Corridor. While based on French SNCF technology, the French engineers who designed it nicknamed it le cochon (the pig) for how heavy it needed to be to meet now-repealed US safety standards and how poorly it performed.
  • Nippon-Sharryo DMU, deployed in Toronto as an airport shuttle and in the San Francisco Bay Area as a regional rail service between Marin and Sonoma counties.
  • The WMATA 7000 Series, built by Kawasaki and included for the sake of comparison.

The model itself is simplified, taking into account just the distance between each station, a single top speed for the whole line, and the performance characteristics of each train.

The modeled train starts at zero seconds, accelerates to either its top speed or the modeled track’s stop speed, then decelerates to a station. It pauses at the station for 30 seconds before accelerating again. If the distance between two stations is too short to allow the train to get up to top speed, it simply accelerates until it decelerates.

Initially, I was concerned that this would be an oversimplification. A train’s acceleration and deceleration aren’t linear, and tracks have curves and variable speed limits (such as the notorious B&P Tunnel on the Penn Line). However, the similarity in performance by our worst-performing trains to the scheduled travel times gives me hope that the results of this model are, while not exact, are fairly close to what a more sophisticated model would create. The full results are available here, but a timetable of the fastest trains against the scheduled times is available in the table below each description.

At the lowest speed, there is not a whole lot of difference between the various DMUs and EMUs aside from our two clunkers, the Nippon-Sharryo DMUs and Acela. While the slowest trains don’t save any time on the fastest scheduled travel times, the fastest can save over 45 minutes off the fastest scheduled trip times to either Frederick (from 1:39 to 0:55) or Martinsburg (from 2:05 to 1:19).

Martinsburg-Washington Union Travel Time

Schedule

79 mph

110 mph

125 mph

Fastest Schedule

2:05

Slowest Schedule

2:14

Class 350 Desiro

1:19

1:09

1:09

Class 395 Javelin

1:21

1:09

1:06

Shinkansen N700

1:21

1:09

1:06

Shinkansen 700

1:24

1:13

1:11

Stadler KISS

1:19

1:06

1:02

BART Trains

1:21

1:20

1:20

Nippon-Sharryo DMU

1:46

1:44

1:44

Acela

1:44

1:38

1:38

WMATA 7000

1:21

1:21

1:21

Stadler FLIRT

1:20

1:10

1:10

Frederick-Washington Union Travel Time

Schedule

79 mph

110 mph

125 mph

Fastest Schedule

1:39

Slowest Schedule

1:51

Class 350 Desiro

1:02

0:55

0:55

Class 395 Javelin

1:04

0:55

0:53

Shinkansen N700

1:04

0:55

0:53

Shinkansen 700

1:07

0:59

0:57

Stadler KISS

1:02

0:52

0:50

BART Trains

1:04

1:03

1:03

Nippon-Sharryo DMU

1:26

1:24

1:24

Acela

1:24

1:20

1:20

WMATA 7000

1:03

1:03

1:03

Stadler FLIRT

1:03

0:56

0:56

Going up a level, to 110 mph, the Stadler trains start to shine. Stadler’s KISS, which is going to be deployed on the Caltrain system in the Bay Area, would shave an hour off travel time to Martinsburg and 56 minutes to Frederick.

Going up once more to the less-realistic level of 125 mph, and we find less of an improvement. Here, Stadler’s KISS still comes out on top with a 53-minute trip from DC to Frederick and 62-minute trip to Martinsburg. Still, every minute sees an improvement in ridership, and with these trip times, train travel becomes much more competitive with driving.

The Camden Line sees less dramatic improvements with new trains and maintenance, but they are still relatively significant. The shortest trip time of 65 minutes from Camden Station to Union Station drops by 20 minutes – almost a third – to 45 minutes at 79 mph. At 110, it drops 7 minutes, and at 125 it drops another 2 minutes, with the KISS again outperforming all comers. Our model’s slowest trains, the Nippon-Sharryo and Acela Express, hardly make a dent even at 125 mph – saving just 5 minutes off the baseline schedule.

Camden Line: Camden-Washington Union Travel Time

Schedule

79 mph

110 mph

125 mph

Fastest Schedule

1:08

Slowest Schedule

1:19

Class 350 Desiro

0:44

0:39

0:39

Class 395 Javelin

0:45

0:40

0:39

Shinkansen N700

0:45

0:40

0:39

Shinkansen 700

0:48

0:43

0:42

Stadler KISS

0:44

0:38

0:36

BART Trains

0:45

0:45

0:45

Nippon-Sharryo DMU

1:02

1:01

1:01

Acela

1:01

0:59

0:59

WMATA 7000

0:44

0:44

0:44

Stadler FLIRT

0:45

0:40

0:40

The Penn Line already runs fast so it needs less of an upgrade, but just changing its trains shaves up to 48 minutes off the trip time from Perryville to 62 minutes from 110 minutes at 125 mph.

Penn Line: Perryville-Washington Union Travel Time

Schedule

79 mph

110 mph

125 mph

Fastest Schedule

1:42

Slowest Schedule

1:58

Class 350 Desiro

1:23

1:11

1:11

Class 395 Javelin

1:25

1:09

1:05

Shinkansen N700

1:25

1:09

1:05

Shinkansen 700

1:27

1:12

1:09

Stadler KISS

1:23

1:07

1:02

BART Trains

1:25

1:24

1:24

Nippon-Sharryo DMU

1:45

1:40

1:40

Acela

1:43

1:33

1:31

WMATA 7000

1:26

1:26

1:26

Stadler FLIRT

1:24

1:12

1:12

Minimum Scheduled Travel Time

Maximum Scheduled Travel Time

Class 350 Desiro (79 mph)

Stadler KISS (110 mph)

Stadler KISS (125 mph)

DC-Martinsburg

2:05

2:14

1:19

1:06

1:02

DC-Frederick

1:39

1:51

0:55

0:47

0:45

DC-Camden

1:05

1:19

0:44

0:38

0:36

DC-Perryville

1:42

1:58

1:23

1:07

1:02

The reason Acela performs so poorly is because it is locomotive-hauled and so has terrible acceleration and deceleration: every stop adds significantly to its schedule. Just by switching to the DMU model, MARC could save a significant amount of time on its schedule without breaking the bank. As mentioned earlier, making this model accurate to the existing speed limits on the ground will reduce the savings somewhat, but it’s clear what a difference the right train makes even on this simplified model.

MARC needs to revamp its plans and buy the right trains. There is no reason besides institutional inertia to maintain a locomotive-hauled fleet. Reducing trip times by a third or more just by switching trains would be an unalloyed win for transit travel in the region.