The Danbury Branch and Rail Modernization
I’ve been asked to talk about how rail modernization programs, like the high-speed rail plan we published at Marron this month, affect the Danbury Branch of the New Haven Line. The proposal barely talks about branch modernization beyond saying that the branches should be electrified; we didn’t have time to write precise branch timetables, which means that the timetable I’m going to post here is going to have more rounding artifacts. The good news is that modernization can be done cheaply, piggybacking on required work on the main of the New Haven Line.
Current conditions
The Danbury Branch is a 38 km single-track unelectrified line, connecting South Norwalk with Danbury making six additional intermediate stops. All stations have high platforms, but they are short, ranging between three and six cars.
Ridership is essentially unidirectional: toward Norwalk and New York in the morning, back north in the afternoon. There is little job concentration near the stations. Within 1 km of Danbury there are only 5,000 jobs per OnTheMap, rising to 10,000 if we include Danbury Hospital, which is barely outside the station’s 1 km radius (but is not easily walkable from it). Merritt 7 is in an office park, but there are only 6,000 jobs there, and nearly everyone drives. The other stations are parking lots, and Bethel is somewhat outside the town center for better parking.
The right-of-way is very curvy, much more so than the main line. Where most of the New Haven Line is built to a standard of 2° curves (radius 873 m), permitting 157 km/h with modern cant and cant deficiency, the Danbury Branch scarcely has a section straight enough with gentler curves than 3°, and much of it has such frequent 4° curves that trains cannot go faster than 100 km/h except for speedups of a few seconds at a time to recover delays.
A first pass on infrastructure and operations
It is effectively free to electrify a 38 km single-track line. The high-speed rail report estimates it at $75 million based on both European electrification costs (see report for sources) and the Southern Transcon proposal, which is $2 million/km on a busy double-track line. The junction between the branch and the main line is flat, but outbound trains can be timetabled to avoid conflict, and inbound trains have no at-grade conflict to begin with. If platform lengthening is desired, then it is a noticeable extra expense; figure $30 million for each eight-car platform, or perhaps half that on single track (but then some stops are double-track), maybe with some pro-rating for existing platforms if they can be easily reused.
The tracks should also be maintained to higher speed, which is a routine application of a track laying machine, with some weekend closures for construction followed by what should be an uninterrupted multidecade period of operations. The curves are already superelevated to a maximum of 5-6″; this is less than the 7″ maximum in US law (180 mm here), but the difference is not massive. The line has a 50 mph speed limit today for the most part, whereas it can be boosted to about 100-110 km/h depending on section, a smaller difference than taking the main line’s 70 mph and turning it into 150-160 km/h.
With a blanket speed limit of 110 km/h – in truth some sections need to dip down to 100 or even less whereas the Bethel-Danbury and Merritt 7-Wilton interstations can be done mostly at 130 – the trip time between South Norwalk and Danbury is, inclusive of 7% pad, 28.75 minutes. The Northeast Corridor report timetables have express New Haven Line commuter trains arriving South Norwalk southbound at :15.25 every 20 minutes and departing northbound at :14.75, so they’d be departing Danbury at :46.5 and arriving :43.5. Meets would occur at the :20, :30, and :40 points.
The :30 point, important as it is a meet even if service is reduced to every 30 minutes, is just south of Branchville, likely too far to use the existing meet at the station. Thus, at first pass, some additional double-tracking is needed, a total of 6 km if it covers the entire Cannondale-Branchville interstation, which would cost around $50 million at MBTA Franklin Line costs. MBTA Franklin Line costs are likely an underestimate, since the terrain on the Cannondale-Branchville interstation is hillier and some additional earthworks would be required on part of the section. A high-end estimate should be the cost of a high-speed rail line without elevated or tunneled segments, around $30 million/km or even less (cut-and-fill isn’t needed as much when the line curves with the topography), say $150 million.
The :20 point southbound is at or just south of Bethel. While this is in a built-up area, the right-of-way looks wide enough for two tracks and the topography is easier; if the station is the meet, then the cost is effectively zero, bundled into a platform lengthening project. Potentially, this could even be further bundled with moving the station slightly south to be closer to the town center. The :40 point southbound is at Merritt 7, which has room for a second track but not necessarily for a platform at it, and could instead get a second track on the opposite side of the platform if there’s enough of a rebuild to turn it into an island with additional vertical circulation; the cost of the second track itself would be a rounding error but the cost of station reconstruction would not be and would likely be in the mid-tens of millions.
How this fits into the broader system
The timetable in the report already assumes that New Haven Line service comprises 6 peak trains per hour (tph) that use the branches. The default assumption, reproduced in the service network graphic, is that New Canaan and Danbury get 3 tph each, and New Canaan gets a grade-separated junction but Danbury does not. Those trains all go to Grand Central with no through-running: only the local trains on the New Haven Line get to run through, since local trains are the highest priority for through-running. If a tunnel connecting the Gateway tunnel with Grand Central is opened, as in some long-term plans (here’s ETA’s, which isn’t very different from past blog posts’), then they can run through to it.
The establishment of this service is not going to, by itself, change the characteristic of ridership on the line. Electrification, better timetabling, and better rolling stock (in this order) can reduce the trip time from an hour today to 29 minutes, and the trip time to Grand Central from about 2:25 to 1:09, but the main effect would be to greatly improve the connectivity of existing users, who’d be driving to the parking lot stations more often, perhaps working from the office more and from home less, or taking the train to social events in the city. Some would opt to use the train to get to work at Stamford, as a secondary market. Over time, I expect that people would buy in the area to commute to work in New York (or at Stamford), but housing permit rates in Fairfield County are low and only limited TOD is likely. It would take concerted commercial TOD at the stations to produce reverse-peak ridership, likely starting with expanding the Merritt 7 office park and making it a bit less auto-oriented.
If the ridership isn’t there, then a train every 20 minutes is not warranted and only a train every 30 minutes should be provided. This reduces the double-track infrastructure requirement but only marginally, as the meets that are no longer needed are the easy ones and the one that still is is the hard one to build, south of Branchville. In effect, something like 80% of the cost provides two thirds of the capacity; this is common to rail projects, in that small cuts in an already optimized budget lead to much larger cuts in benefits, the opposite of what one hopes to achieve when optimizing cuts.
Pedestrian Observations 
Great, readable, easy to follow analysis as always (and hits close to home, as I grew up in the area)!
Curious if you get, or got, engagement from CTDOT or MTA on the Northeast Corridor Report, and if this follow-up was a result of that?
Relatively low Danbury branch ridership is an issue, with limited growth potential. What about the Waterbury branch, with similar length but farther from NYC?
It appears that CTDOT may be thinking about dual mode battery-catenary locomotives for Danbury & Waterbury branches. That might avoid NIMBY reactions to catenary construction, but at a cost. Locomotive service and cost would be worse than electrification, as pointed out in https://www.etany.org/baffling-battery-blunder and https://bqrail.substack.com/p/why-is-the-mta-planning-to-order
Most online maps have property lines on them. On Google maps they are there when you zoom in enough. Very faint but there. Or you can look at the satellite images and see where the fences are. Or you could look at the online tax maps and burrow down enough to get actual dimensions. It probably is, because steam trains spewed… wait for it… steam. And cinders and burning embers and when the railroad was proposed they had to buy something at least 4 rods wide. So the steam etc spewed onto railroad land.
Buy a clue. Well, clues. I suspected it used to be …… slightly south….. at Depot Place. Places like Railroad Avenue, Depot Place, Station Plaza etc. usually have something to do with the trains, a clue! And Wikipedia says it moved north in 1996. And that the 1899 depot is still there. On…. wait for it…. Depot Place. Another clue. I suspect, so that downtown such as it is, didn’t turn into a parking desert. Downtown isn’t as downtown-y if it’s a parking desert. Because it’s Fairfield County Connecticut where “everybody” owns a car or can afford to call a cab. Or both. To the parking desert a few blocks from downtown.
And just what is a car owner in Danbury going to Bethel for? The Bethel CVS instead of the CVS in the mall in Danbury? Someone car free in Bethel, take a long walk from the Danbury station to C-Town and a long cab ride home, with groceries, instead of the independent supermarket in Bethel? There is life north of the Cross Bronx Expressway. It’s somewhat different than Times Square.
What goal does that achieve other than scratching railfan itches?
For the umpteenth time trains can get from the New Haven Line to Penn Station, without railfannery across Midtown. Since 1917. They do it all day every day since…1917. The MTA is even doing something about it. Pity it has taken forever but it doesn’t require billions dollars of pointless tunnels across Midtown.
https://en.wikipedia.org/wiki/Penn_Station_Access
Is the CVS in Larchmont that execrable? Perhaps it’s the lack of Dollar General? Though except for the store brand stuff, CVS likely has the same stuff as Dollar General. So would Shop Rite, using the 66 bus might make more sense for the New Rochelle Shop Rite. And they have a much better selection of food compared to CVS or Dollar General.
I can imagine the railfan splatters when the Stamford local via Sunnyside is on track 9 and the Stamford local via Grand Central is on track 10. Which is what billions of dollars of tunnel across Midtown gets you. You really have to learn how to juggle more than one thing at a time.
Why did London spend billions of dollars building the Elizabeth line? It’s totally pointless and nobody uses it. Same for Paris with the RER. Waste of time.
Because their subways were overcrowded? And slow because they are all-local all-the-time? New York City has express subway trains. And they aren’t all local all the time because they curl up and go to sleep every night.
People from Larchmont have a way to get to Penn Station. They have since 1917. Trains run express through their station today to go there. And run throooooooooooooooughhhhh!! to New Jersey and beyond.
https://pedestrianobservations.com/2025/05/05/the-northeast-corridor-report-is-out
I’ll buy you a clue. People who are on a train to the West Side are not on the overcrowded train to the East Side. They aren’t on the overcrowded shuttle train to the West Side either. All without billions of dollars of tunnels. That clueless railfans think would be a good idea. A different way to look at the clue is that people who want to go to the West Side don’t give a flying leap about the East Side. People who want to go to the East Side don’t give a flying leap about the West Side. And getting to the West Side express – ish is faster than stopping at 125th Street and Grand Central.
This might make your brain meltdown. Someday far in the future, because the MTA has been dawdling for decades, if enough people switch from going to the East Side to get to the West Side they might be able to run another train or two between Grand Central and Harlem line. All without spending billions of dollars on tunnels. That is a complicated concept. Reducing demand through Mount Vernon East ( a New Haven line station) means there can be more trains through Mount Vernon West ( a Harlem line station).
The railfans will be able to get their cheap thrills in Queens where they will be able to transfer from New Haven Line trains to a wide selection of Long Island Railroad trains. or vice versa, without going into Manhattan. I’m sure there will be dozens and dozens of them, normal people, not railfans working up a froth, during busy hours.
Nope, Paris built the RER in order to re-shape the entire metropolitan area. And it worked. The Elizabeth line is having the same effect on the areas it serves (what London needs is two or three more such lines). Through-running regional rail in NY can do the same. For the costs involved it can have incredible benefits to the metro area as a whole.
London has at least 2 RER lines as it already has Thameslink.
They had to reshape the metro area because the Metro was overcrowded and slow. The Elizabeth line hasn’t been open long enough to reshape much of anything except relieve overcrowding on the slow Underground trains. Trips people it take because it’s much faster. Either of them getting express trips. Like the New York Subway has been providing to far flung suburbanizing outer boroughs for over a century. With service on different express trains to the East Side or the West Side.
What part of
was difficult to understand? They can get to Penn Station now, without tunnels from Grand Central.
If someone is on a train that does not go to Grand Central they aren’t contributing to the overcrowding at Grand Central. Whether that is on the commuter trains or the subway. What part of that is difficult to understand?
All without building tunnels. I’m sure you can understand not-building something is cheaper than building something. That they don’t need because they have had a way to get to Penn Station since 1917. Which the MTA will let them use real soon. Did I mention they have been able to do that since 1917 without tunnels from Grand Central to Penn Station?
New York City has 10 and some weird hybrids but since they don’t call them RER railfans ignore them.
@adirondacker12800, aren’t the New York subway express lines really slow?
They are faster than the local. For instance the A train, the 8th Ave. Express, is four minutes faster than the C train, the 8th Ave. Local between 125th Street and 59th Street-Columbus Circle.
The A train, the 8th Ave. line express
https://www.mta.info/document/9466
The C train, the 8th Ave. local.
https://www.mta.info/document/9471
I leave it up to you to make other comparisons. And it seems that quite a few clueless railfans commenting here cannot grasp the concept that if someone is on an express train they cannot simultaneously be on a local train. Someone else can use the space the on the local. Or someone on a train to Penn Station leaving space on a train to Grand Central.
It’s also faster than taking a bus cross town to the Metro North Station on 125th, going to Grand Central and changing to the subway to get back across town and changing subway trains to get to Columbus Circle. Many railfans think rational people would do that.
One thing that’s always baffled me about this line is the lack of station in downtown Norwalk. I wonder how adding a station there impact the rest of the line and what upgrading it would look like (particularly because that section is constrained by a single track tunnel)?
How would you respond to criticisms of the Transit Costs Project’s HSR cost database that say it ignores that the need for stations etc. will vary between projects and that it should have used construction inflation instead of PPP?
1. Stations are included in the projects.
2. Construction inflation makes assumptions on costs escalating over time that aren’t consistently true – for example, relative to the amount of tunneling used, French HSR lines today are built at lower inflation-adjusted cost than the Tokaido Shinkansen was 65 years ago.
No, the criticism wasn’t that stations were excluded, it was that different projects will have different station frequencies and different station scales.
How is that a criticism? Different projects have different station frequencies, but ultimately the end user doesn’t care about costs per station, they care about costs per mile of service.
The point is that since not every project will have the same station frequency or station scale , those projects where it makes sense to have a higher station frequency or a larger station scale are unfairly penalised.