Deinterlining: simpler subway service, fewer delays (New York City)

An approachable explanation from the Effective Transit Alliance of the benefits of deinterlining reverse branches in the New York City subway system. This applies to other train systems too.

This is one of the most complicated subjects about railroad service, but the article explains it well.

Definitions:

• Line: a subway service
• Trunk: a collection of lines sharing tracks for some distance (imagine: tree trunk)
• Simple branch (often deinterlined): line(s) splitting off the trunk (imagine: tree branches)
• Reverse branch (often interlined): when two independent trunks are connected by a shared branch; delays in one cause delays in both

Why it's important: Delays are inefficient. People intuitively don't like transferring between services because the worst-case scenario of transferring increases the probability of at least 1 train in your commute being delayed. Deinterlining alleviates those worst-case scenarios and makes transfers less cumbersome, while also improving the predictability of service for riders who are not transferring. These psychological improvements increase the attractiveness of subway ridership instead of driving cars. A mode shift to transit is more resource-efficient than driving cars, so deinterlining even has overarching financial and environmental benefits even to people in NYC who don't ride the subway.

Computer analogy: Deinterlining is like modularizing concurrent code so components are better encapsulated and don't share resources (threads/queues). They still interface, but a slowdown in one module doesn’t cascade through the entire system. (I'm doing my best for you ~comp people)

The New York City Subway is famously complicated. Many lines are three or four tracks; trains can run local or express; and, crucially, different services often branch and merge with each other. It is this complexity that not only creates the potential for delays, but allows those delays to spread far and wide.

Most well-designed transit networks use branches to fill a central trunk, if branched at all. For example, the A train has branches running to both Far Rockaway and Lefferts Blvd, which allows higher frequency at busier stations in Brooklyn and Manhattan. The same is true of D and N trains, which have separate branches toward Coney Island, but share tracks running express on 4 Av. This type of branching is relatively easy to schedule, as trains can be made to arrive at their merge point at staggered times so that they fit into a neatly spaced pattern in the middle. In this way, trains don't conflict with one another, and they provide more service to denser areas.

The subway, however, also has many reverse branches, where services that have already branched out from one central trunk line join with a branch of another. A prominent example are the 7 Av (2/3) and Lexington Av (4/5) Expresses: these lines branch in the outer boroughs, but the 2 and 5 trains, which run separately in Manhattan, merge onto the same tracks along White Plains Rd in the Bronx and Nostrand Av in Brooklyn. Reverse branches are incredibly difficult to schedule: ensuring that trains always come to interlockings at different times quickly becomes a very difficult, and often impossible, problem to solve. Perhaps worse, when delays mess up this delicate balance, they can cause cascading delays across lines that would otherwise be unaffected.

There is one simple solution to this scheduling and delay-inducing headache: eliminating reverse branches through a process called deinterlining. Deinterlining simplifies the subway’s complex and delicate arrangement, creating a more robust, reliable network. This is why, as the broader transit advocacy has noted for years (e.g., Alon Levy, vanshnookenraggen, Uday Schultz, NYTIP, Joint Transit Association, Mystic Transit), deinterlining is a crucial step in enabling better service on the subway.

The article discusses the "F/M" swap, which is difficult to explain, but in short it reduces delays throughout the entire subway system by switching the tunnels used by these two trains and therefore reducing the number of spatial conflicts with other interlined services.

The article also discusses Nostrand Junction (Rogers) (2+5 and 3+4 conflicts). As with many train service problems, delays from this junction could be resolved with grade-separating the lines: building physically separate tracks so that trains don't have to worry about colliding. However, that's expensive. A nearly-as-good solution is "strategically adding new switches" to deinterline the services. In other words, they can improve the efficiency of the system without extremely costly infrastructure changes.

Lastly, the article discusses the DeKalb Interlocking, a notoriously inefficient part of the subway system. The 6th Avenue "Communications-Based Train Control" (CBTC) project will update the physical systems that control train movements on the B/D/N/Q. Assuming MTA can overcome the logistics of implementing CBTC on multiple sets of rolling stock asynchronously, deinterlining this interlocking would result in meaningful improvements to service and a reduction in delays to the system as a whole, with relatively minor side effects.

Most subway systems are not as complicated as NYC's, so they don't have this problem as much. But for NYC, effective deinterlining could be a huge time savings for the entire system at a fraction of the cost of grade separation.