How to spend $650 million and make things worse as a result...

As you might know, the Northern Indiana Commuter Transportation District (NICTD) finished a major multi-year $650-million double-tracking project about half a year ago.  Currently, the NICTD has only one line – the South Shore line (or the Lakefront Corridor as they started referring to it recently) – although a second line is under construction and should be finished some time in 2026.  Historically, the South Shore line has had four distinct segments with different infrastructure and correspondingly different service levels, decreasing from west to east (from Chicago to South Bend):

1.  Chicago Millennium station – Kensington (quadro-track segment shared with the Metra Electric District line; right-hand running)
2.  Kensington – Gary (double-track; also right-hand running)
3.  Gary – Michigan City (used to be single-track with several passing sidings)
4.  Michigan City – South Bend Airport (single-track with only two sidings)

The segment that has been double-tracked by this program is the formerly single-track segment between Gary and Michigan City.  As a result, the line is now fully double-track from the junction with Metra Electric at Kensington to the 11th Street station in Michigan City (in fact, to a point slightly east of that station, CP 33).

Now, what can go wrong with double-tracking?  Indeed, double-tracking is, in general, a very positive development that increases the rail line's capacity, allows more service, and enhances both reliability and on-time performance.  Or so I thought when I decided to make several trips to Indiana and explore the rebuilt infrastructure.  Luckily, the South Shore line is relatively close to Ann Arbor, and I was able to make four one-day trips to Indiana this fall, some on weekends and some on weekdays (as I did not have to teach during our Fall Break).  I visited all five rebuilt stations, and I made three complete roundtrips over the newly double-tracked segment by train.  I really wanted to like the improvements, and I was coming to Indiana with a very positive outlook.

And some improvements are indeed positive.  The rebuilt stations are nice.  The construction quality is reasonably good.  Most new high-level platforms are long (8 cars) and have a nice surface texture preventing slipping.  Aesthetically, many elements are pleasant and likable, although not extraordinary.  They have also slightly increased service levels east of Gary on weekdays (but are still running only 9 roundtrips a day on weekends).  I might post some pictures later on, but here is a track map of the rebuilt segment that I drew as a result of my extensive exploration. Station names on this map are:
GMC = Gary Metro Center
M = Miller
P/OD = Portage/Ogden Dunes
DP = Dune Park
BS = Beverly Shores
11th = 11th Street (Michigan City)
CA = Carroll Avenue (Michigan City)

So, why do I say that things got worse as a result?  Ultimately, most problems I describe below can be traced to one singular very poor design decision at the project's planning and engineering stage.  Namely, at two out of the five rebuilt stations (Portage/Ogden Dunes and Dune Park), they built high-level platforms on only one track.  Moreover, they built them on the southern track, which would have been an eastbound (outbound) track for normal right-hand operations.  The platforms on the northern (westbound/inbound) track are still low at these two stations.

However, it is clear – and someone at the NICTD understands this – that high-level platforms are more important for boarding than for alighting.  Climbing up is harder than descending for most people.  So, the way the NICTD now runs their trains is by routing westbound (inbound) trains to the "wrong" track at these two stations in order to serve the high-level platforms.  In essence, they try to operate the segment through Portage/Ogden Dunes and Dune Park as a single-track segment with all trains serving the high-level platforms on the southern track.

The problem is that this does not work well – or at all.  Even during off-peak hours or on weekends, the effectively single-track segment is an operational bottleneck.  But during rush hours it is simply impossible to operate both inbound and outbound traffic on this single track reliably.  So, what does the NICTD do?  Well, they did not come up with any better ideas but to run trains on the left-hand side through this area.

Normally, most double-track lines in the U.S. and in the world are operated with a fixed track assignment: trains run on one track in one direction, and trains in the opposite direction run on the other track.  Exceptions are hard to find.  Now, on the South Shore line, the western end of the line in Chicago is right-hand operated.  The NICTD cannot change it, because it is owned and run by Metra, and Metra's operation on their entire line is exclusively on the right-hand side (which makes perfectly logical sense for Metra).  Yet a segment further east is left-hand operated by the NICTD during busiest hours.  In essence, this creates conflicting moves for each and every train on the South Shore line.  Moreover, it creates these conflicting moves twice, since the NICTD operates the newly double-tracked segment through Michigan City on the right-hand side as well.  Just think about it: you get a left-hand operated segment sandwiched between two right-hand operated ones!  You get conflicts on both sides of the left-hand operated segment, with every single train having to switch tracks twice per trip.

What are the consequences of this "slalom"?  First of all, the on-time performance is absolutely the worst in several decades.  Most trains are delayed 10-15-20 minutes, and some are delayed severely.  One of the trains I rode on October 26 was delayed by an hour.  Again, think about it: a commuter train with the entire route well under 90 minutes is delayed by an hour!  This is not a transcontinental multiday trip on Amtrak – this is a commuter operation – and such delays are totally unacceptable for such a time-sensitive operation.

Moreover, most of their trains I took are severely crowded, even on weekends, and they cannot add more trains because of these conflicting moves twice per each trip restricting their ability to add more trips.  While many modern double-track lines can accommodate over a hundred roundtrips a day, the NICTD is struggling with pitiful two dozen (in fact, even fewer than two dozen).  And all of that is because they saved probably 1% or 2% of the project budget on these two platforms on the northern track...

The negative consequences do not end here.  Trains switch from one track to the other at seemingly random locations, and these locations vary day by day.  It feels like their dispatchers are improvising every time instead of following a predetermined operational plan.  So, even the exact same train may be arriving at the exact same station on a different track.  At the same time, there is no real-time information for passengers at any of the rebuilt stations.  Passengers are not informed at all on which track the next train will arrive.  Even though they have these numerous expensive electronic signs, they only display scheduled information, not the real-time one, and they don't have any information whatsoever on which track the next train is running on.  There are no permanent signs and no audio announcements either.  As a result, passengers are frantically running back and forth from platform to platform without any understanding as to where to expect their train (and when to expect it).  Over the four days that I spent exploring this line, I was asked at least three times "At which platform do you think the next train to Chicago is going to stop?"  And of course, every rebuilt station has separate platforms on each track – another poor design decision, – and you have to cross at least one track in order to get from one platform to the other (see the track diagram for the platform layout at each station).  So, instead of improving things for passengers (who used to have only one platform to wait on without any ambiguity), double-tracking has made it worse for them.

I have seen probably a hundred commuter-rail operations all over the U.S. and in 30 other countries, and I can declare with absolute certainty that right now the South Shore line is the most hectic, disorganized, and unpredictable commuter rail operation that I have ever seen anywhere in the world.  The NICTD's response?  It's all leaves!  No, it is not the incompetence of whatever consulting company they used to design the "one-sided" stations, nor the incompetence of their managers or dispatchers, no, it's all just leaves.  And of course, it is the very first autumn in over a century, and leaves have not fallen on their tracks ever before, so they have never encountered such an amazingly rare event in their entire lives, and they have no idea whatsoever how to deal with it! :-)  "Pathetic" does not even begin to describe it...

And this is not the end of this story.  Unfortunately, design and engineering deficiencies are not limited to the two "one-sided" stations.  The rebuilt 11th Street station has two high-level platforms (which is a good thing), but only one of them is long enough for 8-car trains, and the other one can only fit 4 cars.  In essence, they now have to route long trains to only one platform on only one track.  Why?  There was enough space to build the second full-length platform.  Why would you intentionally handicap your operations in such a silly way?  Who on earth makes such poor design decisions?  Moreover, the full-length 8-car platform is on the northern track at this station.  It means that if they route all long inbound trains to a high-level platform at each station, they have to operate on the right side through Michigan City (as they do now), and hence they have to switch between the two main tracks at least twice (first between Michigan City and Dune Park and then again between Portage/Ogden Dunes and Kensington).  In essence, the "slalom" is forced by the newly built infrastructure.

Another example: The new Miller station has two full-length high-level platforms, but the two stub tracks for short turns and train storage are connected to only one main track and hence can serve only one boarding platform (and only one stub track is electrified).  It means that all short turns can only use one track at Miller, during all hours of operation, which, in its turn, means that most through trains have to use the other track at Miller, in both directions at all times, and hence one cannot assign well defined time periods of right-hand running and left-hand running through Miller.  Each and every one of these poor design decisions results in operational constraints and inefficiencies, with trains constantly needing to switch from one track to the other unnecessarily.

As you can see, the newly built infrastructure imposes a lot of limitations, and these are mostly self-inflicted limitations resulting from horrible decisions made during this double-tracking project.  These infrastructure constraints cannot be easily fixed or undone by competent scheduling.

It seems that their cost-cutting was very severe, and it is seen everywhere.  All electrification poles used on the newly double-tracked segment are made of wood, and only wooden cross-ties are employed everywhere on the newly laid track.  No sane person would use wood for such items in the 21st century.  If you look around, concrete for ties and metal for poles are the materials of the day.  Wood does not last long and needs to be replaced far too soon.  In essence, the NICTD insists on building a 19th-century railroad in the 21st century.  At the same time, in the visually nice Miller station the seats are hard and made of metal.  No cushions.  Apparently they think that commuter rail passengers do not deserve cushioned seats, unlike airport passengers.  I would invite the NICTD's CEO to spend an hour waiting for their delayed train on one of such metal seats and then share his experience.

Now, how can they fix this mess?  For one thing, the NICTD needs to build second full-length high-level platforms at Portage/Ogden Dunes and Dune Park to restore consistent right-hand operations throughout the entire line.  This will help with both on-time performance (as all conflicting moves will be eliminated) and passenger information (as they will now be able to place permanent signs "inbound" and "outbound" on each platform).  Again, this is how most double-track lines in the world are operated: one track in one direction, and the other track in the other direction, – and this line should not be an exception.  It should be possible to apply for grants and receive such grants relatively easily, as it is straightforward to understand why such "one-sided" stations are an operational nightmare.

In the short term, they need to hire a software engineer and make sure that real-time information is displayed on those electronic signs at stations, both the arrival time and the arrival track, so that passengers stop running back and forth between platforms across active tracks.

Moreover, the NICTD needs to stop and rethink the second line they are building right now.  I am sure there are similarly poor decisions made for this line as well.  I am aware of at least one: They decided not to electrify a short segment of that line.  Instead, they will buy much more expensive rolling stock that can operate both from electricity and from diesel.  Why on earth would one want the incompatible rolling stock for their two lines?  Who needs a zoo with different maintenance needs and incompatible maintenance practices?  It is both cheaper and more efficient in the long run to simply electrify this short segment and use the same rolling stock of EMUs throughout their entire system.  The unparalleled genius of the NICTD decision makers does not stop at bad decisions – they seem to pursue the absolutely worst ones!

Ann Arbor Mini-High

An Amtrak train next to the mini-high platform at the Ann Arbor station.

The Ann Arbor station is located on the east-west Detroit – Chicago line in Michigan (USA) and serves Amtrak Wolverine trains between these cities. The station has a rather unique mini-high platform. Over the years several different people asked me for photos or descriptions of this platform. As a result, I decided to write a comprehensive post to illustrate the concept and to avoid haphazard attempts to describe it in the future. Photos were taken at different times in different years (2017-2021), so small details may differ slightly, but the big picture stays the same throughout.

First, a view of the station from a nearby overpass is intended to provide an idea of the overall layout of the station: a long low-level platform to the east and a short mini-high platform to the west.

The Ann Arbor station. From the European standpoint, it is not really a station,
just a stop, as it has only one track and no switches.

Next, a closer look at the mini-high platform. The high-level platform is referred to as "mini-high" because it is rather short – it is only long enough for a couple of adjacent doors in two adjacent cars.

A view of the mini-high from the station's parking lot.

A view of the mini-high from the track side.
The white portion in the middle is retractable and can move in or out.

Another view from the track side.
There are ramps and stairs to the low-level portions at both ends.

A view of the mini-high from the low-level portion.

Note how far the mini-high is set back from the track compared to the low-level portion. This is necessary to provide sufficient clearance for freight trains with extra-wide loads.

Another view from the other end.

One of the two ramps leading up to the high-level portion and
providing full accessibility for disabled riders and for riders with luggage or prams.

A view from the high-level platform.

The portion with the yellow tactile edge on the above photo can move to the right towards the track. The metal gates in the very front and in the very back can swing open to provide access to this movable portion once it is in its boarding position, as will be seen below.

Another illustration of how far from the track the movable portion is in its retracted position.

When a train arrives at the station, it needs to be positioned in such a way
that its two adjacent doors are next to the movable portion of the mini-high.

Once the train is in position, the movable portion extends to the train under control of Amtrak
employees and eliminates the gap, thus providing level boarding/disembarking for all passengers.

Note that the side railings of the movable portion also move with it, thus preventing passengers from potentially falling while boarding. Also note that the swinging gates are now open, providing access to the movable portion. Once the boarding process is complete, the movable portion is retracted back to its original position.

The gap between a moving Amtrak train and the mini-high platform is pretty large – almost a meter!

While most photos were taken during COVID intentionally avoiding large crowds and may produce an impression of lack of passengers, Ann Arbor is generally a busy station – in fact, the busiest in Michigan – with 100-150 passengers typically boarding each train on weekends and during busy travel periods (holidays, beginning/end of each semester, etc.) Since the introduction of the mini-high platform, boarding time for all these passengers has decreased dramatically from more than 10 minutes to just a few. Everyone benefits from faster boardings, even passengers boarding at other stations, with trains having shorter running times. To achieve the same time savings by improving the rail line itself, one would need to invest significantly more than the cost of this mini-high. Thus, a mini-high platform is a prudent investment in speeding up trains and keeping them on time. I hope this successful improvement will be replicated at other busy stations with low platforms where a full-length high-level platform cannot be built.

“Netgraphs”

Here is a very interesting graphical way to represent timetable information for a regularly running service, whose clockface schedule repeats every hour (but not necessarily at equal time intervals between trains). The authors of the idea called it a “netgraph”. To illustrate this idea, take a look at the netgraph of Zurich S-Bahn or the netgraph for the entire Swiss rail network (the latter has an English version of the legend).

In essence, each solid line represents a train per hour running every hour. Different colors represent different types of trains (on the netgraph of Switzerland) or different S-Bahn routes (on the netgraph of Zurich S-Bahn). If a train runs every two hours, it is represented by a dashed line. If a train runs during rush hours only or makes just a few trips per day, it is represented by a dash-dot line. The numbers near each station represent the train’s arrival time from a particular direction (the one closer to the station rectangle) and the train’s departure time to the same direction (the one further from the station rectangle) in minutes after each hour. You can find the other (less important) notations in the legend.

As a result, it becomes possible to display all the timetable information for the entire network on a single page. I find this idea fascinating and promising. It makes it so easy to grasp general tendencies of the timetable and so simple to use it for both passengers and operators. For example, it is extremely easy to see all possible transfers at a given station, and how long each transfer takes (the time between the arrival of one train and the departure of another). That is, one can perceive the timetable not at the “one train at a time” level or even the “one route or line at a time” level, but as “the entire network at a time”.

Of course, such “netgraphs” require regular clockface timetables repeating every hour, like they do in Switzerland. There are many other benefits of having a regular timetable with coordinated/timed transfers (such as the ability to remember the timetable for the lines you use frequently), but this simplified way to represent it is an additional benefit that should not be overlooked.

If you’d like to see more netgraphs, there are several others (Luxemburg, Basel, etc.) here.

Railjets

The locomotive end of a Railjet trainset in the Austrian colors at the Graz station.

Last winter, VIA Rail Canada ordered 32 new trainsets from Siemens for its Windsor – Toronto – Ottawa – Montreal – Quebec City corridor. These trainsets will completely replace the current rolling stock on this line. The new trainsets will be almost identical to the Railjet trainsets operated by the Austrian and Czech railroads throughout Central Europe and also produced by Siemens. I had a chance to ride Railjets a couple dozen times last summer, so here is more information about these trainsets.

The cab-car end of a Railjet trainset in the Czech colors at Vienna’s Main station.

Railjets are operated in the push-pull mode, and there is a locomotive at one end and a cab car at the other. While the locomotive can be decoupled and replaced easily, all cars are semi-permanently coupled in each trainset, with hermetically insulated gangways between them. Train orientation is always fixed and same: The locomotive always pulls when going westward through Vienna and always pushes when going eastward through Vienna. While European Railjets are hauled by the electric Siemens Taurus locos, VIA Rail trainsets will be hauled by the diesel Siemens Charger locomotives similar to those throughout Midwest.

Semi-permanently coupled cars with an insulated gangway between them. Note the retractable step at the bottom.

Each European Railjet has 7 cars, one of which is a cab car. Typically, there is one first class car, one café/restaurant car, and five second class cars (known as “coaches” in the U.S.). The trainsets ordered by VIA Rail will have a smaller number of cars: only 5 cars each. Thus, the overall number of Siemens cars ordered by VIA is 160, almost double the Amtrak Midwest order. The length of each car is 26.5 m (87 feet).

A seven-car Railjet trainset in the Austrian colors at the Graz station.

A seven-car Railjet trainset in the Czech colors at the Graz station.

The European trainsets can be operated in a formation of two (i.e. 2 engines and 14 cars), reaching the overall length of 410 m (more than a quarter mile). Importantly, station infrastructure allows operation of such long trains, and platform length exceeds 410 m at all major stations. Train doors are automatic, and all doors open at all stations, speeding up alighting and boarding significantly. Two trainsets can be coupled and decoupled en route easily, which is done on regular basis at major junction stations in under 10 minutes.

Signs at the end of a platform in St. Pölten indicating the cab stop of 410-meter trains.
The entire platform is more than a quarter mile long.

The trainsets are rated for the maximal speed of 230 km/h (143 mph). I have personally ridden a Railjet in passenger operation at 220 km/h (137 mph) on the Westbahn line in Austria. The power output of each electric Taurus loco is 6.4 MW (8600 hp), i.e. almost double that of each diesel Charger loco in Midwest (4400 hp). Other technical characteristics of Railjets can be found here.

The locomotive end of a Railjet trainset in the Austrian colors at the Graz station.

The cab-car end of a Railjet trainset in the Austrian colors at the Vöcklabruck station.

The locomotive end of a Railjet trainset in the Austrian colors at the Villach station.

The cab-car end of a Railjet trainset in the Austrian colors at the Graz station.

The locomotive end of a Railjet trainset in the Austrian colors at the Villach station.

The vast majority of European Railjets are owned by the Austrian Railways (ÖBB), and several are owned by the Czech Railways (ČD). Austrian trainsets and locomotives are colored in bright red and dark red, while Czech trainsets and locomotives are in light blue and dark blue, but all are technically similar.

The cab-car end of a Railjet trainset in the Czech colors at Vienna’s Main station.

A Taurus locomotive in the Czech colors at Vienna’s Main station.

The cab-car end of a Railjet trainset in the Czech colors at Prague’s Main station.

Some ÖBB trainsets are in darker colors. The locomotive end of a Railjet trainset in darker colors at the Salzburg station.

The cab-car end of an ÖBB Railjet trainset in darker colors at the Linz station.

All long-distance services in Austria and many international routes to neighboring countries are now operated with Railjets, for example:
Munich – Salzburg – Vienna – Budapest
Zurich – Innsbruck – Salzburg – Vienna – Vienna’s Airport
Prague – Brno – Vienna – Graz
Vienna – Villach – Venice
Few trains terminate at Vienna’s Main station. Instead, most operate through the station, taking advantage of the through tracks at the new Main station completed in 2015. This is extremely good for capacity, as each train takes track space at the station for only 5-10 minutes, and the station is not used as a train parking lot. As a result, Vienna’s Main station manages to accommodate more passenger trains per hour than the entire State of Michigan per week.

Passenger rail routes operated with Railjets. Full size (3524 × 3444).

Notably, Railjet cars have excellent seats: not too hard, not too soft, at just the right height. As a result, travel in Railjets takes much lesser toll on one’s back than travel in Amtrak seats, which are too soft and too low above the floor. When purchasing a ticket, one can reserve a specific seat for only 3 euros. The seat diagram provided on the website allows one to select the seat they prefer: window or aisle, forward facing or backward facing, on the left side of the train or on its right side. Since train orientation is fixed, one is guaranteed the desired orientation of their seat as well. Over the two dozen trips I made on Railjets last summer, not a single train had a different orientation from the one advertised during seat purchase. This ability to reserve a specific seat is a far cry from the humiliating and degrading need to stand in a long line in order to get a seat of choice on an Amtrak train.

The interior of a Czech second-class quiet car.

The interior of an Austrian second-class quiet car.

Each trainset has a quiet car in its second-class portion. I observed on a few occasions that conductors actually maintain the quiet environment. Once a conductor pointed out to two teenagers, who argued relatively loudly, that this was a quiet car. A few times conductors asked someone talking on the phone to hang up or to move to another car. As a result, the ride in a quiet car was actually quiet, even in an almost full car.

The interior of a first-class car.

Overall, Railjets are excellent trainsets, modern and very pleasant. I think VIA Rail made a prudent decision to go ahead with a proven European design, especially such a good one. I can attest that these are some of the best passenger rail cars available on the market, and I am looking forward to riding these trainsets on the Canadian Corridor.

Update 1: Railjet service between Prague and Berlin indicated on the above map began on June 15, 2020.

Update 2: Amtrak ordered 83 similar integrated trainsets from Siemens in summer 2021. They will be operated on the Northeast Corridor and various state-supported routes in the Northeast as well as on the Cascades. The number of cars per each trainset has not been reported yet.

Atlanta's rail system

About a month ago I visited Atlanta. Atlanta's MARTA remained the only metro / heavy rail system in the U.S. that I did not explore in any detail or even see up until that point. So, I decided to use three days of our spring break to have a little rail vacation and rectify this very last omission.

MARTA

Atlanta's MARTA is a modern metrorail system similar to those in Washington DC and San Francisco Bay Area (BART). It was established at about the same time and was built to the same modern standards: spacious stations, wide platforms and passageways, fast running trains (definitely above 100 km/h), larger distances between stations, large park-and-ride lots at outer stations, etc. Some observations:

Positives:

 ‑ Trains are very fast. A very pleasant experience. Track is mostly in good condition, although I remember a couple of rough stretches.

 ‑ While some stations begin showing their age, overall, they are in good condition and well maintained.

 ‑ Stations have many entrances/exits and good pedestrian distribution systems at each exit, which are well designed. This is in stark contrast with systems like Los Angeles, where "a single whole in the ground" seems to be the norm.

 ‑ The system is clean (at least by relatively low American standards). In fact, I was shocked to see cleaning employees at many stations throughout the system, who were actually cleaning the system, for example, wiping escalators' handbelts with some sanitary solution. This is not a common sight in most U.S. systems. Cleanliness is definitely above average. No bad smells or physiological fluids, even in bad parts of town.

 ‑ Despite Atlanta's notorious crime reputation, the system felt safe in all of its parts.

 ‑ People (passengers) are better behaving that in some older systems, particularly, New York or Chicago. No one was playing music without headphones (a big problem in New York City), no one was eating on the trains, no one was yelling all the time. Pretty pleasant.

Negatives:

 ‑ The system features extremely low ridership. During the day, I would sometimes find myself to be the only passenger in a rail car or the only person in a rail station. I think a big part of the problem is that MARTA has completely decimated its bus system. There are very few feeder bus routes with any decent frequency. You can get only so much ridership from people walking to your stations or parking at your stations. Feeder bus network is extremely important, and Atlanta gets very low marks on such a feeder network. Bus terminals at many rail stations sit underused or even completely/partially closed. This is not a good feature of the transit system. They need to increase bus service frequency, this would help to build rail ridership and fill their running trains. The current bus "system" hurts the rail system's ridership.

 ‑ The rail system has an extraordinary number of unnecessary audio announcements on trains. Customers are constantly and continuously bombarded by audio pollution. And these are not just station announcements (which are also made too frequently, in my opinion), there are many totally useless announcements as well. For example: "Everyone would like to enjoy a pleasant ride. Please review MARTA's code of conduct at itsmatra.com." Seriously? Does one really need to hear a meaningless announcement of this sort in two languages every 5 minutes? What is its informational content? And there are many others, equally senseless and meaningless announcements. This rail system gets nominated to be the worst in the U.S. in terms of the extreme number of announcements. In essence, the loudspeakers are talking almost continuously. Cut it down! They need to reduce the number and the frequency of announcements drastically. There is absolutely no justification for subjecting MARTA's passengers to such a horrendous amount of unnecessary noise.

 ‑ Blue line had delays on all three days. They seemed to be unable to keep the schedule on this line throughout my visit. On a couple of occasions, I had to wait for the train for almost half an hour! I have no idea if there was a good reason, but with a train every 12 minutes, there is little excuse for such poor schedule adherence.

 ‑ While the daytime frequency is quite adequate (6 minutes on the trunk lines and 12 minutes on branches, further reduced to 5 and 10 minutes respectively during rush hours), the night and weekend frequency is very poor: a train once in 20 minutes. Especially at night, when Red and Green lines are converted to shuttles and do not provide direct service to the main transfer point (Five Points) in Downtown Atlanta, it is virtually impossible to get anywhere in a reasonable amount of time. Around 9 pm service frequency on the busier north-south line drops from "every 6 minutes" down to "every 20 minutes". This is unacceptable. I witnessed a young couple entering a station and seeing sign "the next train is in 18 minutes" and asking each other: "Are they serious?" Indeed, MARTA cannot be serious with such frequencies. This is pathetic for an urban rail system.

 ‑ The busiest branch seems to be the northern end of the Red line. Ridership on this branch is healthiest and heaviest. This can be observed by a naked eye, but I am sure there is data to support this assertion. Yet, this is precisely the line that gets cut short during evening hours. Why? Why the heck do they run night shuttles on this line? Why not to cut short the less busy Gold line? Is anybody thinking? At night, Red and Gold service patterns need to be swapped.

At the very least, I would make the Red line shuttle run all the way to Five Points in the evenings. This section between Five Points and Lindbergh Center is the busiest section on the entire system, and there are turnaround facilities just south of Five Points, so this seems to be a no-brainer. This would make the system infinitely more convenient for practical use in the evenings and at night. And it would not even cost that much: they would need just one extra Red line train, given their current 20-minute night-time intervals on all branches.

Peculiarities:

 ‑ The Peachtree Center station in the crown jewel of this rail system. The station was carved in bedrock and has bare walls, well, made of rock. It looks stunning. It also has a well developed system of exits to several points throughout Downtown. This is the only deep station on this rail system. A "must see".

 ‑ All platforms are 180 m long and can fit 8-car trains, but only 6-car trains are running on the three main lines (Red, Gold, and Blue). However, there is one exception: the platform at the Bankhead station is only long enough for two cars. This is the main reason why the Green line operates with two-car trains at all times. Longer trains would simply not fit at the Bankhead station's platform. This seems to be rather shortsighted. I have no idea why this station was built in such a substandard way, but they will need to rectify this shortcoming sooner or later.

Conclusions:

Overall, MARTA is a good system. However, there is room for improvement. My top three recommendations:

1. Drastically reduce the number and the frequency of in-train announcements. This costs nothing, but makes a huge difference for passengers.

2. Extend the night shuttle on the Red line to the Five Points station and thus reduce service frequency on the busiest portion of the system between Five Points and Lindbergh Center from "once in 20 minutes" to "once in 10 minutes". This will cost very little, as they will need to run only one more train and have one more operator in service during evening/night hours.

In addition, swap night service patterns on the Red and Gold lines. Ideally, the Red line should run all the way south to the airport, and the Gold line should terminate at Five Points (not at Lindbergh Center).

3. Improve the bus feeder network. A bus once an hour is a ridiculous frequency for a populated and growing city like Atlanta. Run them every 15 minutes or less and be surprised with a sudden surge of ridership. Houston did this just a couple of years ago, follow their suite. This is higher cost, but still much cheaper than building new lines or stations.

Atlanta streetcar

I really don't want to offend anyone here. But my softest, nicest expression about this rail system is this: What a pathetic joke! The streetcar is very slow. It runs in street traffic and gets stuck in traffic. There are no published timetables. Streetcars run empty. No fare enforcement. No fare integration with MARTA. Only one-way connection to MARTA. The streetcar was not running one day I was in Atlanta, but no announcements, fare machines are operational and still sell 2-hour and 1-day tickets, even though there will be no streetcars until the next day (yes, I wasted a dollar). I found only one positive: nice vehicles and nice seats with low-back support. Everything else is pretty much mediocre. The grade is C.

There are also two airport people movers in the Atlanta airport (which is the largest in the U.S.), one in the secure zone and one outside, but they are really no different from other similar airport systems, so I will save my breath here.

Overall, my conclusion is that Atlanta is worth a two or three day visit for a railfan.

Capacity and reliability improvements in South-East Michigan

Michigan Department of Transportation (MDOT) has completed two important rail projects in the Metro Detroit area this winter. One is the so-called West Detroit Connection, and the other is the second main-line track between the Wayne Junction and the Dearborn station. Both projects will affect passenger rail service in the region.

The West Detroit Connection (WDC) is a single-track rail connector between the east-west Conrail main line (co-owned by Norfolk Southern and CSX) and the north-south Canadian National (CN) main line (map). The newly built overpass structure is owned by MDOT, while the connecting track is now owned by CN. As another part of this project, a mile (1.6 km) of the second track has been added to the CN main line north of the connector, as well as several crossovers.

The main purpose of this connection is to provide a shorter, more direct route for passenger trains that bypasses two busy control points and avoids interference with freight traffic on the Conrail line. Even though the reduction in travel distance is not very significant (less than a mile), the new route is expected to improve reliability and on-time performance by reducing both the number and the extent of potential delays, as it avoids two interlockings where freight congestion can hold up passenger trains. The connector is designed for the top speed of 45 mph (72 km/h) due to its curvature; however, the main time savings will be realized from the increased reliability due to the lack of conflicts rather than high speed. This logical improvement has been needed ever since the Detroit station was moved from the old Michigan Central location to its current spot in the New Center area in 1994, and it is finally open for service more than 20 years later. Amtrak trains have been using the new connection on daily basis since December.

As a result of construction of the connector and the second track on the CN main line, two rail junctions have been modified at each end of the new double-track segment: CP West Detroit and CP Vinewood. My track schematics of the new layouts at both locations can be found here (page 1 is CP West Detroit, and page 2 is CP Vinewood). Although the connector itself is single-track, Amtrak trains can now access both main tracks at both ends of the connector, thanks to the new crossovers installed on both CN and Conrail lines at CP West Detroit. The schematics also illustrate the elimination of conflicts between Amtrak trains and NS/CSX freight trains. The former move from Dearborn to Detroit (or in the opposite direction) via the newly built connector, while the latter move from the Livernois Yard or Toledo to Canada or Detroit (or in the opposite direction) via CP Scotten on the Conrail territory, and there is no single point of conflict between these movements any more.

The second rail project completed by MDOT this winter is a 9-mile (14-km) segment of the second main track between the Wayne Junction and the new Dearborn station on the former Michigan Central line (a part of the Wolverine Corridor currently owned by MDOT). All track work was finished before the end of last year, but the new track is not in service yet. According to MDOT, it is waiting for the Positive Train Control signal work to be completed before it can be placed in service. The new track has been constructed to the north of the existing track. Several crossovers have been added, and all grade crossings have been rebuilt as a part of this project.

The segments west of the Wayne Junction and east of the Dearborn station have been double-track for many years. Completion of the new second track between them creates a continuous double-track line from CP Ypsilanti to the West Detroit Connector, a distance of 25 miles (40 km). Moreover, since the single-track WDC is very short (about a quarter of a mile, or 400 m), and since a mile of second track has been added on the CN main line north of the WDC, almost all 28 miles (45 km) between CP Ypsilanti and the Detroit New Center station are now double-track, with the only exception of the connector. This expanded capacity paves the way for the future Ann Arbor – Detroit commuter rail, another overdue project for South-East Michigan. Double-tracking the segment between the Wayne Junction and Dearborn was also one of the stipulations of the sale of the former Michigan Central line to MDOT by Norfolk Southern, which retains freight trackage rights on this line.

In other news, both platforms are currently in use at the new Dearborn Amtrak station. The northern platform has been in service since December 2014, while the southern platform was not used until recently. During meets at this station, westbound trains typically stop at the northern platform, and eastbound trains now stop at the southern platform.

However, the gate between the southern platform and the Henry Ford Museum remains permanently closed, preventing easy pedestrian access to this major attraction and defying the concept of an “intermodal” station, more than a year after the station’s opening. Amtrak passengers are currently advised to take a more-than-a-mile-long roundabout taxi trip instead of a short direct walk between the platform and the museum. It is hoped that this deficiency will be rectified soon and the gate will eventually be opened and begin serving Amtrak passengers, as originally intended.

Other infrastructure improvements along the Wolverine Corridor will continue this year and next. All projects are scheduled to be completed by the end of 2017, and higher-speed passenger service is expected to be initiated on this line shortly thereafter.

Three new rail stations opening this spring

Three new major intercity rail stations are opening in the U.S. this spring. The three stations are in Denver, CO, St. Paul, MN, and Miami, FL.

Denver Union station (map) reopened for rail service on February 28, when Amtrak began using the new station. Amtrak was using a temporary platform for several years while construction of the modern multi-track facility proceeded. The new station features 8 tracks: 2 tracks with low-level platforms for Amtrak and tourist/special trains and 6 tracks with high-level platforms for future regional/commuter trains, which will begin serving four lines radiating from Union station in 2016. The track layout of the new station is as follows (the dead-ends are on the left):

PLATFORM
(PLATFORM) Track 8 (Regional Rail - High Level)
Track 7 (Regional Rail - High Level)
(PLATFORM) Track 6 (Regional Rail - High Level)
PLATFORM
Track 5 (Amtrak - Low Level)
SERVICE PLATFORM
Track 4 (Amtrak - Low Level)
PLATFORM
(PLATFORM) Track 3 (Regional Rail - High Level)
Track 2 (Regional Rail - High Level)
Track 1 (Regional Rail - High Level)
PLATFORM
STATION BUILDING

There is also an underground bus transfer facility with 22 bus bays, which will be used by various RTD bus routes and will replace the existing Market Street transfer center a few blocks away. The bus portion of the station is opening on May 9, and the renovated station building is opening on July 11. Thus, by the middle of this year Denver Union station will be in full service - except there will be no trains (apart from two Amtrak trains a day) until 2016, when three of the four regional rail lines (including the line to the airport) are expected to open. Denver is implementing one of the most ambitious transit expansion programs in the nation, and so far they seem to be on track and on schedule.

One of the major design flaws of the new station is in the inconvenience of future transfers between regional rail and light rail. The light rail station has been moved two city blocks (200 m) away from the regional rail/Amtrak station. Being a terminal dead-end station, it would have been very simple to build the light rail stop near the end of the stub tracks, so that transferring passengers would have only a short walk to get to the connecting train. Yet the light rail station was moved to the furthest corner of the development, thus maximizing the inconvenience of rail-to-rail transfers.

Another new rail station opening soon in the U.S. is the renovated St. Paul Union Depot (map). The present "Midway" Amtrak station will be taken out of passenger service, and all Amtrak operations will be moved to Union Depot. Overall, it is a good idea, because public transit connections at Union Depot are so much better, while at the current location they are simply non-existent. The stop at Union Depot will be more convenient not only for St. Paul, but also for Minneapolis, because there will be a direct light rail line from Union Depot to Downtown Minneapolis beginning this June. And it will also be possible to get to the soon-to-be-closed Midway station by taking the same light rail line to the rough vicinity of the current station. I see no loss in abandoning the old station.

The new St. Paul station will not be as large as the one in Denver. There will be only two tracks and one low-level island platform between them. Passengers will not have to cross any tracks though - there is vertical access from the new platform to the Depot building above, which is already a great improvement over many other station designs on the Amtrak system.

The Union Depot project is delayed by more than a year now, and it is currently estimated that Amtrak will be moving to the new station in early May - and hopefully before the light-rail Green line begins serving Union Depot in June.

Finally, yet another major rail station opening soon is the Miami Central Station next to the current terminus of the Metrorail Orange line in Miami International Airport (map). The new dead-end station is expected to open for the Tri-Rail regional rail service in June. It will be connected to the airport terminal by the MIA People Mover. The track layout includes four tracks and two low-level island platforms, which will be arranged as follows:

Track
PLATFORM
Track
Track
PLATFORM
Track

Just like in Denver, this is going to be a pretty major station by the U.S. standards, as very few rail stations have more than only one or two tracks with boarding platforms.

Amtrak will not move to the new Miami Central Station until 2015 though. Station design did not take into account the length of the locomotive and the "slack" distance needed to trip the grade crossing signals. As a result, if a long Amtrak train is stopped at the new station, it will be blocking a street at the north end of the station. Florida DOT will now have to spend $25 million to build a "bypass" road, so that automobiles could use the next grade crossing a few blocks north while an Amtrak train is blocking the grade crossing next to the station. Intelligent, isn't it?

Update 1: St. Paul Union Depot reopened for train service and the Amtrak station moved to Union Depot on May 7, 2014.

Update 2: After long delays, Miami Central Station opened for the Tri-Rail regional rail service on April 5, 2015. Amtrak service to this station is now expected to begin in 2016.

System maps

A system map is a complete map of a transit system, which contains all rail and bus routes within the service area of a transit agency as well as information on service hours and service frequency for all services.  A system map is perhaps the most important informational product a public transit agency produces.  It contains the most complete information about services provided by the agency on a single piece of paper, allowing users to have this complete information at their fingertips at any time before and during a trip by public transit.  A system map is an indispensible travel tool to navigate a transit system unsurpassed by any other informational products offered by transit agencies.

In North America (and I mean the U.S. and Canada here) the situation with system maps has been very favorable over the last decades:  Most large agencies produce system maps of rather high quality, and folded versions of these maps are freely available to riders at transit stations and in customer service centers.  Typically, one can also request a system map from a transit agency by email or phone, and they gladly mail a copy upon such a request in most cases.

Lately, however, an alarming trend has developed among some transit agencies:  they simply discontinued production of paper versions of their systems maps.  They still produce their maps and update them, they still spend their money on the most difficult job - to keep the map current, - it is just they do not print them anymore, referring customers to online versions.  There are several large agencies that have adopted such a policy; to name a few, King Country Metro (Seattle, WA), TransLink (Vancouver, BC), Port Authority of Allegheny County (Pittsburgh, PA), and RIPTA (Providence, RI) are on this list.

Some agencies went even further and stopped production of system maps altogether.  The most notorious example here is perhaps New Jersey Transit, which did not produce a map of their bus system in more than a decade, not even an electronic version.  Other agencies limit distribution of their printed maps; for example, Denver RTD will not mail their system maps outside the state of Colorado.

This tectonic shift in policy on system maps seems to be caused by several factors.  The first (and perhaps most important) factor seems to be the lack of understanding by transit managers of the importance of system maps as the principal informational product of their agencies.  Since most transit managers do not use public transit in everyday life, few of them seem to understand how important a good system map is as an actual planning tool for their riders, both existing and new ones.  Some have even suggested that there is no significant value in system maps outside of an academic interest in the breadth and coverage density of transit networks.  As a life-long transit user, I navigated my way by transit with system maps through many cities in North America and in Europe, and availability of a good system map is often what makes a difference between a good transit system and a mediocre one.  In essence, a system map provides all the information one needs about public transit in a geographic area, and if this information is not available, one’s usage of the transit system is severely impeded.

The second reason for discontinuing paper maps is often provided along the financial lines: “Due to funding shortage...” or “Due to dwindling resources…” etc.  What I do not understand, however, is why the agencies cannot simply SELL these paper maps for a small fee, instead of distributing them for free, like some agencies have been doing for quite a while.  For example, Philadelphia's SEPTA produces ones of the best transit maps I have ever seen; it is not just a distorted network of bus lines, but a full city map with all the streets and all the transit information, yet they don't distribute those maps for free - they sell them for a few dollars apiece.  Similarly, Sacramento RTD sells its system maps at 25 cents each.  Many European agencies also sell their maps, for example, Berlin’s BVG.  Such an approach makes paper maps available for those who need them, while covering the costs of printing and storage.  It also deters waste, because people are more reluctant to dispose of something they paid money for (even if it is only a couple of bucks).  The agencies listed above should adopt a policy of selling the maps instead of discontinuing them altogether.

The third factor influencing the system maps policy is related to a persistent perception that an online version somehow replaces a paper map, and hence publication of paper maps is not such a priority or even necessity.  I certainly understand the general trend towards everything digital, and this trend has brought us mostly good things so far.  For example, availability of transit schedules in digital format online offers a great example of how the new technology simplifies transit usage.  Online schedules can be accessed using smartphones and other mobile devices in real time or printed in advance, as long as they are in a printable (PDF) format rather than on some hard-to-print flash-enabled webpages.  The main problem with digital content is specific to maps, especially online system maps.  It is really the map that needs to be on paper.  A user can print letter-sized PDF schedules of the needed bus routes, but one typically does not have the means to print a meter-by-meter system map.

A typical misconception goes along these lines:  “There is a dearth of smartphones and mobile devices available nowadays, and plenty of applications allow a customer to pan and zoom through either a digitized map or a data layer to get the specific detail, as well as to locate stops and to generate trip plans with directions.”  To begin with, the problem with online maps is precisely that: they are online, not in one’s pocket or backpack.  There are still plenty of spots in any city where neither Internet nor cell-phone connection is available, most notably in major parts of their rail transit systems - in the underground tunnels where many urban rail lines run.  Not everyone owns an Internet-enabled smartphone yet, so it is also an issue of equity.  These are rather obvious initial arguments against discontinuance of paper maps and over-reliance on online maps.

But more fundamental arguments run much deeper.  To address the above misconception, we need to distinguish two different scenarios of mobile-device usage for transit planning purposes:  (A) looking at an online map to plan one’s transit trips by oneself, and (B) letting an application or a “trip planner” to plan a trip for a user.  These are not the same, and I will consider these scenarios separately.

(A)  When it comes to planning one’s trip by oneself by just looking at a map, what most people do not seem to realize is that most mobile devices are simply too small to show a large system map in any meaningful way.  Simply nothing beats a meter-by-meter paper map when it comes to finding one’s way in a big city.  Viewing a large map on a 5-inch screen is similar to a torture.   There is no digital device on the market yet, which would have a meter-by-meter display, which could be folded and stored in a pocket, which would weigh less than an ounce, and which would be capable of accessing online maps even on a train deep underground.  And BEFORE such a device is created, any discussion of online maps as a replacement of paper maps is completely meaningless, simply because it is NOT an equivalent replacement.  No amount of “zooming and panning” will even compare to a good paper map, just because one DOES NOT NEED to zoom and pan when planning one’s trip on a paper map - all the information is in one’s sight without the limitations of the tiny screen of one’s favorite electronic gadget.

(B)  Letting an app or a “trip planner” do the planning is a completely different story.  Of course, if these apps were done perfectly, that would be a magical solution: let the computer do its thing for you!  However, there are severe limitations here as well.

Perhaps most importantly, all of these apps and trip planners produce an itinerary, which is only good for a specific day and time.  They might give one an itinerary for Friday, 4:30 pm, which will not work on Sunday or at night, simply because a bus route it suggested runs only during rush hours.  Thus, one has to plan his or her trip EACH TIME when using these trip planners.  None of them can possibly produce an advice like this: “On weekdays before 9 pm take bus A, but at nights and on weekends take bus B and then transfer to bus C”.  While looking at the map, I can get such a general idea very easily, all I need to know is my route options and service hours, both of which are typically available on the single system map without any typing, zooming, or panning.

Yet another disadvantage of various automated trip planners is that they do not have any idea about such notions as “frequency” and “reliability”.  Most (if not all) of them plan a trip based on theoretical transit schedules.  But what if you missed your connection, because the first bus was late?  If the second line runs every 5 minutes, it is not such a big deal, but if it runs once an hour, it is a serious problem, which can delay you enormously.  None of the trip planners will make a determination that it is possibly safer to take a frequent reliable line rather than a non-frequent unreliable one.

Connection times are frequently estimated based on some very crude assumptions, and in my experience most trip planners produce an exaggerated walking time, based on the speed of a very slow walker, such as an elderly or disabled person.  It is not uncommon to see a 10-minute prediction for a transfer where the actual walking time by a healthy person is only 2 minutes.  This overestimation discourages walking and transfers between different lines and favors direct rides, even if they are actually slower.

The apps are made by people, and people set their priorities in their programs.  Of course, some of the more intelligent “trip planners” allow one to choose “fastest trip”, “shortest walk”, “least transfers”, or “shortest wait”.  But the problem is: this is still a machine, and if one tells it “walk no more than 500 m”, it will cut all the options that include more walk - even if it is only 510 m, and one would not even mind such a slightly longer walk.  So, the main point, computers have no common sense, and an option that might be acceptable to you (and might even be the best) would not be shown.  And vice versa: computers may be selecting some non-sense options, just because they have no intelligence.

Many of these apps are severely flawed.  I will provide two examples.  I will never forget standing on the inbound platform of the Woodside LIRR station (in Queens) last year, when I was approached by a teenage girl with a smartphone in her hand.  The girl asked me if trains to Penn station depart from this platform (the answer was “yes”) and when the next train is coming (“in 3 minutes”).  Then I asked “But where are you going to?”  She said: “Great Neck station”.  I asked “What Great Neck station?”, and she showed me her smartphone: it was the Great Neck LIRR station on the Port Washington line.  For those, who are not familiar with the LIRR system, the Port Washington line also stops at the Woodside station (where we were standing), but in order to get to the Great Neck station, the girl would have to take a train in the opposite direction from the outbound platform.  In essence, the app advised the girl to travel to Penn station and then transfer there to a Port Washington train instead of taking a Port Washington train directly from the Woodside station without going to Penn station and back.  I do not know what company programmed that app and what options the girl selected (the longest trip? the highest fare?), but she certainly got a silly itinerary that a simple look at the system map would have never produced.

As another example, I have just tried to plan a trip by the New York MTA Trip Planner for a midday weekday trip from JFK to a station on the B line in Brooklyn that I made three weeks ago.  Even with all the best options I could possibly set and select, it gave me three itineraries, none of which included the route I actually took, and every one of which had a travel time that was longer than the trip I actually made.

Additionally, Google Transit is known to have outdated schedules, as was the case recently, when I tried to plan a trip in the Boston area on a Saturday, and it offered me to ride a bus route that no longer runs on weekends.  I knew it was discontinued on July 1, 2012 because I follow transit issues in that area, but there would be no way of knowing this for someone relying on Google entirely.

In general, I am very skeptical about these “ready solutions” in the form of the apps that plan a trip for you.  In most cases, I can do a much better job by myself, and so can other people.  These applications may indeed be a perfect solution for people who cannot read a map; however, there is no need to dumb the rest of us down to their level and make us use these "ready solutions" by discontinuing paper maps.

At the same time, I do not oppose transit agencies spending taxpayer money for developing such electronic applications.  What I oppose is taking away a good product - the printed system maps - and replacing it with a mediocre one.  These electronic gadgets are still nowhere near printed map in terms of their utility and convenience yet.

Paper maps should stay, and every transit agency should continue publishing them at regular intervals.