Penn Station 3D Model
As part of our high-speed rail program at Marron, I designed and other people made a 3D model of the train station I referenced in 2015 in what was originally a trollish proposal, upgraded to something more serious. For now there’s still a password: letsredothis. This is a playable level, so have a look around.
The playable 3D model shows what Penn Station could look like if it were rebuilt from the ground up, based on best industry practices. It is deliberately minimalistic: a train station is an interface between the train and the city it serves, and therefore its primary goal is to get passengers between the street or the subway and the platform as efficiently as possible. But minimalism should not be conflated with either architectural plainness (see below on technical limitations) or poor passenger convenience. The open design means that pedestrian circulation for passengers would be dramatically improved over today’s infamously cramped passageways.
Much of the design for this station is inspired by modern European train stations, including Berlin Hauptbahnhof (opened 2006), the under-construction Stuttgart 21 (scheduled to open 2025), and the reconstruction of Utrecht Central (2013-16); Utrecht, in turn, was inspired by the design of Shinagawa in Tokyo.
As we investigate which infrastructure projects are required for a high-speed rail program in the Northeast, we will evaluate the place of this station as well. Besides intangible benefits explained below in background, there are considerable tangible benefits in faster egress from the train to the street.
Moreover, the process that led to this blueprint and model can be reused elsewhere. In particular, as we explain in the section on pedestrian circulation, elements of the platform design should be used for the construction of subway stations on some lines under consideration in New York and other American cities, to minimize both construction costs and wasted time for passengers to navigate underground corridors. In that sense, this model can be viewed not just as a proposal for Penn Station, but also as an appendix to our report on construction costs.
Background
New York Penn Station is unpopular among users, and has been since the current station opened in 1968 (“One entered the city like a God; one scuttles in now like a rat” -Vincent Scully). From time to time, proposals for rebuilding the station along a better or grander design have been floated, usually in connection with a plan for improving the track level below.
Right now, such a track-level improvement is beginning construction, in the form of the Gateway Project and its Hudson Tunnel Project (HTP). The purpose of HTP is to add two new tracks’ worth of rail capacity from New Jersey to Penn Station; currently, there are only two mainline tracks under the Hudson, the North River Tunnels (NRT), with a peak throughput of 24 trains per hour across Amtrak’s intercity trains and New Jersey Transit’s (NJT) commuter trains, and very high crowding levels on the eve of the pandemic; 24 trains per hour is usually the limit of mainline rail, with higher figures only available on more self-contained systems. In contrast, going east of Penn Station, there are four East River Tunnel (ERT) tracks to Long Island and the Northeast Corridor, with a pre-corona peak throughput of not 48 trains per hour but only about 40.
Gateway is a broader project than HTP, including additional elements on both the New Jersey and Manhattan sides. Whereas HTP has recently been funded, with a budget of $14-16 billion, the total projected cost of Gateway is $50 billion, largely unfunded, of which $20 billion comprises improvements and additions to Penn Station, most of which are completely unnecessary.
Those additions include the $7 billion Penn Reconstruction and the $13 billion Penn Expansion. Penn Reconstruction is a laundry list of improvements to the existing Penn Station, including 29 new staircases and escalators from the platforms to the concourses, additional concourse space, total reconstruction of the upper concourse to simplify the layout, and new entrances from the street to the station. It’s not a bad project, but the cost is disproportionate to the benefits. Penn Expansion would build upon it and condemn the block south of the station, the so-called Block 780, to excavate new tracks; it is a complete waste of money even before it has been funded, as scarce planner resources are spent on it.
The 3D model as depicted should be thought of as an alternative form of Penn Reconstruction, for what is likely a similar cost. It bakes in assumptions on service, as detailed below, that assume both commuter and intercity trains run efficiently and in a coordinated manner.
Station description
The station in the model is fully daylit, with no obstruction above the platforms. There are eight wide platforms and 16 tracks, down from 11 platforms and 21 tracks today. The station box is bounded by 7th Avenue, 31st Street, 8th Avenue, and 33rd Street, as today; also as today, the central platforms continue well to the west of 8th Avenue, using the existing Moynihan Train Hall. No expansion of the footprint is required. The existing track 1 (the southernmost) becomes the new track 1A and the existing track 21 becomes the new track 8B.
The removal of three platforms and five tracks and some additional track-level work combine to make the remaining platforms 11.5 meters wide each, compared with a range of 9-10 meters at some comparable high-throughput stations, such as Tokyo.
With wide platforms, the platforms themselves can be part of the station. A persistent difference between American and European train stations is that at American stations, even beloved ones like Grand Central, the station is near where the tracks are, whereas in Europe, the station is where the tracks are. Grand Central has a majestic waiting hall, but the tracks and platforms themselves are in cramped, dank areas with low ceilings and poor lighting. The 3D model, in contrast, integrated the tracks into the station structure: the model includes concessions below most escalator and stair banks, which could offer retail, fast food, or coffee. Ticketing machines can be placed throughout the complex, on the platforms as well as at places along the access corridors that are not needed for rush hour pedestrian circulation. This, more than anything, explains the minimalistic design, with no concourses: concourses are not required when there is direct access between the street and the platforms.
For circulation, there are two walkways, labeled East and West Walkways; these may be thought of as 7⅓th and 7⅔th Avenues, respectively. West End Corridor is kept, as is the circulation space under 33rd Street connecting West End Corridor and points east, currently part of the station concourse. A new north-south corridor called East End Corridor appears between the station and 7th Avenue, with access to the 1/2/3 trains.
What about Madison Square Garden?
Currently, Penn Station is effectively in the basement of Madison Square Garden (MSG) and Two Penn Plaza. Both buildings need to come down to build this vision.
MSG has come under attack recently for competing for space with the train station; going back to the early 2010s, plans for rebuilding Penn Station to have direct sunlight have assumed that MSG should move somewhere else, and this month, City Council voted to extend MSG’s permit by only five years and not the expected 10, in effect creating a five-year clock for a plan to daylight Penn Station. There have been recent plans to move MSG, such as the Vishaan Chakrabarti vision for Penn Station; the 3D model could be viewed as the rail engineering answer to that architecture-centric vision.
Two Penn Plaza is a 150,000 m^2 skyscraper, in a city where developers can build a replacement for $900 million in 2018 prices.
The complete removal of both buildings makes work on Penn Station vastly simpler. The station is replete with columns, obstructing sight lines, taking up space between tracks, and constraining all changes. The 3D model’s blueprint takes care to respect column placement west of 8th Avenue, where the columns are sparser and it’s possible to design tracks around them, but it is not possible to do so between 7th and 8th Avenues. Conversely, with the columns removed, it is not hard to daylight the station.
Station operations
The operating model at this station is based on consistency and simplicity. Every train has a consistent platform to use. Thus, passengers would be able to know their track number months in advance, just as in Japan and much of Europe, train reservations already include the track number at the station. The scramble passengers face at Penn Station today, waiting to see their train’s track number posted minutes in advance and then rushing to the platform, would be eliminated.
Each approach track thus splits into two tracks flanking the same platform. This is the same design used at Stuttgart 21 and Berlin Hauptbahnhof: if a last-minute change in track assignment is needed, it can be guaranteed to face the same platform, limiting passenger confusion. At each platform, numbered south to north as today, the A track is to the south of the B track, but the trains on the two tracks would be serving the same line and coming from and going to the same approach track. This way, a train can enter the A track at a station while the previous train is still departing the B track, which provides higher capacity.
The labels on the signage are by destination:
- Platform 1: eastbound trains from the HTP, eventually going to a through-tunnel to Grand Central
- Platform 2: westbound trains to the HTP, connecting from Grand Central
- Platform 3: eastbound trains from the preexisting North River Tunnels (NRT) to the existing East River Tunnels (ERT) under 32nd Street
- Platform 4: eastbound intercity trains using the NRT and ERT under 32nd Street
- Platform 5: westbound intercity trains using the NRT and ERT under 32nd Street
- Platform 6: westbound trains from the ERT under 32nd Street to the NRT
- Platform 7: eastbound trains to the ERT under 33rd Street and the LIRR, eventually connecting to a through-tunnel from the Hudson Line
- Platform 8: westbound trains from LIRR via the ERT under 33rd Street, eventually going to a through-tunnel to the Hudson Line
Signage labels except for the intercity platforms 4 and 5 state the name of the commuter railway that the trains would go to. Thus, a train from Trenton to Stamford running via the Northeast Corridor and the under-construction Penn Station Access line would use platform 3, and is labeled as Metro-North, as it goes toward Metro-North territory; the same train going back, using platform 6, is labeled as New Jersey Transit, as it goes toward New Jersey.
Such through-running is obligatory for efficient station operations. There are many good reasons to run through, which are described in detail in a forthcoming document by the Effective Transit Alliance. But for one point about efficiency, it takes a train a minimum of 10 minutes to turn at a train station and change direction in the United States, and this is after much optimization (Penn Station’s current users believe they need 18-22 minutes to turn). In contrast, a through-train can unload at even an extremely busy station like Penn in not much more than a minute; the narrow platforms of today’s station could clear a full rush hour train in emergency conditions today in about 3-4 minutes, and the wide platforms of the 3D model could do so in about 1.5 minutes in emergencies and less in regular operations.
Supporting infrastructure assumptions
The assumption for the model is that the HTP is a done deal; it was recently federally funded, in a way that is said to be difficult to repeal in the future in the event of a change in government. The HTP tunnel is slated to open in 2035; the current timetable is that full operations can only begin in 2038 after a three-year closure of NRT infrastructure for long-term repairs, but in fact those repairs can be done in weekend windows—indeed, present-day rail timetables through the NRT assume that one track is out for a 55-hour period each weekend, but investigative reporting has shown that Amtrak takes advantage of this outage only once every three months. If repairs are done every weekend, then it will be possible to refurbish the tunnels by 2035, for full four-track operations in 12 years.
The HTP approach to Penn Station assumes that trains from the tunnel would veer south, eventually to tracks to be excavated out of Block 780 for $13 billion. However, nothing in the current design of the tunnel forces tracks to veer so far south to Penn Expansion. There is room, respecting the support columns west of 8th Avenue, to connect the HTP approach to the new platforms 1 and 2, or for that matter to present-day tracks 1-5.
It is also assumed that Penn Station Access (PSA) is completed; the project’s current timeline is that it will open in 2026, offering Metro-North service from the New Haven Line to Penn Station. As soon as PSA opens, trains should run through to New Jersey, for the higher efficiency mentioned above.
The additional pieces of major infrastructure required for this vision are a tunnel from Penn Station to Grand Central, and an Empire Connection realignment.
The Penn Station-Grand Central connection (from platforms 1 and 2) has been discussed for at least 20 years, but not acted upon, since it would force coordination between New Jersey Transit and Metro-North. Such a connection would offer riders at both systems the choice between either Manhattan station—and the choice would be on the same train, whereas on the LIRR, the same choice offered by East Side Access cuts the frequency to each terminal in half, which has angered Long Island commuters.
Overall, it would be a tunnel of about 2 km without stations. It would require some mining under the corner of Penn 11, the building east of 7th Avenue between 31st and 32nd Street, but only to the same extent that was already done in the 1900s to build the ERT under 32nd Street. Subsequently, the tunnel would nimbly weave between older tunnels, using an aggressive 4% grade with modern electric trainsets (the subway even climbs 5.4% out of a station at Manhattan Bridge, whereas this would descend 4% from a station). The cost should be on the order of hundreds of millions of dollars, not billions—the billions of dollars in per-km cost in New York today are driven by station construction rather than tunnels, and by poor project delivery methods that can be changed to better ones.
The Empire Connection realignment is a shorter tunnel, but in a more constrained environment. Today, Amtrak trains connect between Penn Station and Upstate New York via the existing connection, going in tunnel under Riverside Park until it joins the tracks of the current Hudson Line in Spuyten Duyvil. Plans for electrifying the connection and using it for commuter rail exist but are not yet funded; these should be reactivated, since otherwise there’s nowhere for trains from the 33rd Street ERT to run through to the west.
It is necessary to realign the last few hundred meters of the Empire Connection. The current alignment is single-track and connects to more southerly parts of the station, rather than to the optimal location at the northern end. This is a short tunnel (perhaps 500 meters) without stations, but the need to go under an active railyard complicates construction. That said, this too should cost on the order of hundreds of millions of dollars, not billions.
Finally, platforms 3-6 all feed the same approach tracks on both sides, but in principle they could be separated into two. There are occasional long-term high-cost plans to fully separate out intercity rail tracks from commuter tracks even in New York, with dedicated tunnels all the way. The model does not assume that such plans are actualized, but if they are, then there is room to connect the new high-speed rail approach tunnel to platforms 4 and 5 at both ends.
Overall, the model gives the station just 20 turnouts, down from hundreds today. This is a more radical version of the redesign of Utrecht Station in the 2010s, which removed pass-through tracks, simplified the design, and reduced the number of turnouts from 200 to 70, in order to make the system more reliable; turnouts are failure-prone, and should be installed only when needed based on current or anticipated train movements.
Pedestrian circulation
The station in the model has very high pedestrian throughput. The maximum capacities are 100 passengers/minute on a wide escalator, 49 per minute per meter of staircase width, and 82 per minute per meter of walkway width. A full 12-car commuter train has about 1,800 passengers; the vertical access points—a minimum of seven up escalators, five 2.7 meter wide staircases, and three elevators per platform—can clear these in about 80 seconds. In the imperfect conditions of rush hour service or emergency evacuation, this is doable in about 90 seconds. A 16-car intercity train has fewer passengers, since all passengers are required to have a seat, and thus they can evacuate even faster in emergency conditions.
Not only is the throughput high but also the latency is low. At the current Penn Station, it can take six minutes just to get between a vertical access point and an exit, if the passenger gets off at the wrong part of the platform. In contrast, with the modeled station, the wide platforms make it easier for passengers to choose the right exit, and connect to a street corner or subway entrance within a maximum of about three minutes for able-bodied adults.
This has implications for station design more generally. At the Transit Costs Project, we have repeatedly heard from American interviewees that subway stations have to have full-length mezzanines for the purposes of fire evacuation, based on NFPA 130. In fact, NFPA 130 requires evacuation in four minutes of throughput, and in six minutes when starting from the most remote point on the platform; at a train station where trains are expected to run every 2-2.5 minutes at rush hour and unload most of their passengers in regular service, it is dead letter.
Thus, elements of the platform design can be copied and pasted into subway expansion programs with little change. A subway station could have vertical circulation at both ends of the platform as portrayed at any of the combined staircase and escalator banks, with wider staircases if there’s no need for passengers to walk around them. No mezzanine is required, nor complex passageways: any train up to the size of the largest New York City Subway trains could satisfy the four-minute rule with a 10-meter island platform (albeit barely for 10-car lettered lines).
Technical limitations and architecture
The model is designed around interactivity and playability. This has forced us to make some artistic compromises, compared with what one sees in 3D architectural renderings that are not interactive. To run on an average home machine, the design has had to reduce its polygon count and limit the detail of renderings that are far from the camera position.
For the same reason, the level shows the exterior of Moynihan Station as an anchor, but not the other buildings across from the station at 31st Street, 33rd Street, or 7th Avenue.
In reality, both East and West Walkways would be more architecturally notable than as they are depicted in the level. Our depiction was inspired by walkways above convention centers and airport terminals, but in reality, if this vision is built, then the walkways should be able to support themselves without relying too much on the tracks. Designs with massive columns flanking each elevator are possible, but so are designs with arches, through-arches, or tied arches, the latter two options avoiding all structural dependence on the track level.
Some more architectural elements could be included in an actual design based on this model, which could not be easily modeled in an interactive environment. The platforms certainly must have shelter from the elements, which could be simple roofs over the uncovered parts of the platform, or large glass panels spanning from 31st to 33rd Street, or even a glass dome large enough to enclose the walkways.
Finally, some extra features could be added. For example, there could be more vertical circulation between 7th Avenue and East End Corridor (which is largely a subway access corridor) than just two elevators—there could be stairs and escalators as well. There is also a lot of dead space as the tracks taper from the main of the station to the access tunnels, which could be used for back office space, ticket offices, additional concessions, or even some east-west walkways functioning as 31.5th and 32.5th Streets.
Pedestrian Observations 
The political challenge with this is that New York, and the US in general, is obsessed with defensive architecture. It’s hard to see a plan like this carried out if the state cannot keep homeless people off of the tracks. Is Germany as concerned about this? If so, how does the station work?
To a lesser extent, Amtrak’s (awful) boarding processes serve the purpose of keeping people without Amtrak tickets off the train. At stations like Newark, the conductors spend much of their time yelling that NJT tickets will not be honored on the train. At 30th St or Penn, much less so.
So, how can a plan like yours square with Amtrak’s, and New York’s, goal of making sure that the Wrong Kind of People don’t get on the platform or get on the train?
New Haven has open boarding and it’s fine.
Boarding here is, as you might expect, open. Any bum off the street can get on any train. If it’s an intercity train, a conductor will kick them off if they don’t have a valid ticket, unless they argue successfully (Jon Worth managed to do it in France when SNCF was at fault for some hilariously bad ticket distribution practices on international trips).
All new stations should have platform screen doors on all platforms. Replace the trains if the doors are not in the right place. (I’ve been asking for a few standard trains specifications which state where the doors are allowed to be, and how high the floor is for a while here). We have the technology now and it completely eliminates the homeless person is on the tracks problem.
If you are worried about homeless on the platforms we also have fare gates which are common. There are pros and cons to them vs roving fare inspectors, but they are overall cheap enough to implement. They do need more space and a slightly different station design, so Alon might need to redo their model (I haven’t look at it yet) if we decide this is our future. Or you can just hire some guards to wonder the station – guards can do things like help tourists people figure out which ticket they need to buy, assist a disabled person, and other tasks that happen all the time in a busy station so you need them anyway.
Sorry for my ignorance, I haven’t been to that many train stations outside New York. But the ones I recall at least from Paris (Gare du Nord, Gare de Lyon, Gare de L’est, if I recall correctly) all seemed to have the cafes and news stands in a single location connected to the platforms— certainly closer to the trains than places like Grand Central, yet still closer to each other than the tracks, so that a person looking for food could see all the options. Would it not be better for passengers and the businesses alike for all the shops to be in one location, so that people don’t have to go up and down staircases and from their waiting area (the tracks) just to even see what the options are?
Regardless, thank you for all you are doing in pushing for Penn Station to be built in a positive way with best rail practices! I’d hope for some sort of architecture above that pays homage to the original station, but nonetheless the emphasis must be on the trains, not simply the architecture of the station
Absolutely not.
People going to the train station primarily for its transportation value have no need to actually see all their options, they need to see at least some of their options without taking a major detour. A major train station is a facility that is hundreds of meters long and tens to hundreds of meters wide, so forcing a detour could easily add a half kilometer to kilometer of extra walking.
It should be easy to access the station from many different points from the surrounding streets, and amenities should be easily available no matter where on the surrounding neighborhood you are coming from, instead of all clustered together, far from the tracks, and potentially quite far from the surrounding neighborhood.
There is value in having a cluster of options in one place for those who want to make the detour, which is why having malls above/below/adjacent to the station is great. However, even people who don’t detour to those facilities should have options. And even with station adjacent malls, a large station would benefit from having malls on all sides, not just at one end.
It would be better for some people, but worse for most. The majority of people are not in the station for the shopping at all, and so shops that stand in the way of getting to their train in anyway (just having to walk farther because the shop exists) is a negative.
The shops on the platforms are useful because everyone will eventually wake up late and rush to the station, then realize they have a couple minutes to spare so they will head to the nearest shop to buy whatever they sell for breakfast (assuming the line/wait is short). These people wouldn’t dare risk going to a shop on a different platform or even a shop on the entry way as they need to see their train and thus have confidence that if they misjudge time they can still get on their train (some places you must be on time)
For people who always eat at the station they will already know what their favorite place to eat is. Their routines will be setup to go to whatever platform it is on, get their meal and then move on. These people will find you by exploring the station when their previous regular shop doesn’t meet their needs (closes).
Shops all in one place are a great idea if people are going there for the shops – they can see all their options at once. If it is a family they can send each person to a different shop and then meet up. However this is a completely different business model from a station, stations need to be built for their primary model first: getting people around the city, and then conveniences shoved in where there is extra space. In fact your station should avoid becoming the mall destination itself because then you are competing with the malls for customers and so the malls will want to do more to encourage arriving by car (once someone is on transit they won’t visit other malls since the station is a good enough mall).
This model is fun to “walk” around, but it is hard to orient myself in terms of the street layout, or to see things like subway connections. Do you have a top-down map or some other rendering that provides an overall sense of scale?
Which level would the Penn to GCT enter Grand Central Madison at. You do realize that down near 37/38th St the 4 tracks merge into 2 south of the ventilation facility at 38th ST. Also there’s a TBM buried on the lower level south of 37th St. So connecting into ESA would not just require a short 2km tunnel but would require some quite extensive reconfiguration of the existing 38th St Vent facility, it would require upgrading the tail tracks from storage to housing passengers which means adding bench walks etc. revising the signaling system amongst other things all of which would add to the cost then there’s the need to disrupt LIRR operation while your making the connections. All possible but the cost starts to add up. Then you have the small matter of DEP water tunnels to get around. We looked at this during ESA construction and when Gateway was called ARC.
If you think your getting into MNR’s trainshed, forget it. There’s Park Ave tunnel that gets in the way, together with Grand Central Station and the Steinway tunnel and the 38th ST vent plant shafts. There’s simply no space to run up the west side of Park Ave unless you go beneath the buildings, where easement acquisition costs would blow you budget out of the water. You would have to come in on the east side of Park Ave. Plus which platform tracks would become through running? Lower level GCT is not deep enough to get under the subway and Park Ave tunnel. So there is no way to get into MNR territory at GCT coming in from the south. There’s simply too many existing facilities in the way, unless your planning to build a tunnel between MNR and LIRR in the 35ft of rock that still exists between the cavern crown and the lower level of GCT? Is it possible, sure, but to even get into the lower level of GCT you would need to rebuild NYCT Grand Central Station, rebuild the dining concourse of GCT, relocate the Shuttle, relocate Park Ave Tunnel so that simple little 2km of tunnel just became a massive project. Even with 4% grades this is pure fantasy.
One possible option that we discussed with MNR many years ago would be to bring new tunnels from way up north, under the existing Park Ave box, and then come in to the east side of GCT and create some kind of “subway ” style station for two lines and then continue out on the east side of Park Ave before swinging west to head towards what was then planned to be the 34th St folly that was to be the new ARC station. It was still difficult to get the access for the stations up to the street with all of the overbuild, but this had the advantage of being able to be profiled to get under all the Lex/Park Ave stuff and not be constrained by the existing infrastructure. Of course this required easement acquisition south of GCT for the reasons noted above as you run south out of the existing DOT RoW.
And your rebuild of Penn Station, how exactly are you planning to rip down the overbuild, remove the columns while maintaining a functioning station with the current capacity. Or are you planning to massively reduce capacity in Penn Station just when Gateway and Penn Station Access are finished…… if you want to improve Penn Station, and believe I’m all for a new station, then you might need Penn South to provide the capacity while your closing down chunks of the existing Penn Station to rebuild it. IN nobody’s world can you rip out railroad infrastructure, build new platforms and radically change the track layout without causing massive disruption to operations. Or are you thinking that the railroads will be fine with a 50% reduction in capacity and revenue while you build the new station.
It would require some mining under the corner of Penn 11, the building east of 7th Avenue between 31st and 32nd Street, but only to the same extent that was already done in the 1900s to build the ERT under 32nd Street.
If the building hasn’t been built yet, you don’t have to mine under it. The tunnels opened in 1910, Vornado, the owner of the building, says it was built in 1923.
Tearing down the Hotel Pennsylvania, between 32nd and 33rd included, or did, I’m not going to ferret out the final compromises, was to restore the passageway between Penn Station and Herald Square. There was a passageway between Penn Station and the Hudson and Manhattan ( today’s PATH ), under 33rd. It was there to allow step free access for passengers changing between the PRR, Erie and Delaware Lackawanna and Western and their porter, via the Hudson and Manhattan. I don’t see access to the passageway or access to 32nd Street and the entrance to the IRT/1,2 and 3 trains, that is on32nd. ( The subway station plan on Google maps is accurate as far as I remember. )
…. this should be a new thread.
Well of course its much easier to put the rail in an open cut and then build over it. Its very much more expensive and difficult however to excavate immediately beneath the bottom of an existing building without having to underpin the building as you go, unless of course you have around a diameter, in this case 23ft of Manhattan schist between your tunnel and the underside of the building, and even that might be pushing it with the size of some of these buildings and their loads.
I can appreciate the thought here that all of this would be a great idea, but the practicalities behind it and hence the cost leave a little to be desired.
all true. That doesn’t change that they didn’t have to mine under a building, built in 1923, when they were laying track in 1910 or earlier.
Ad rebuilding Penn Station: remove the above-ground structures (this is why MSG only has a five-year operating permit), build the walkways, move the vertical circulation to the walkways, remove the concourses, start replacing platforms. The existing Penn Station could part with a significant fraction of its platforms right now, and might even be able to part with some after Gateway opens; tracks 1-5 don’t really need to be used (NJT trains should be through-running to PSA), so the new platforms can be built south to north.
There’s no 50% reduction in capacity – the railroads’ stated dwell times are ridiculous. One of the planning docs asserts that Platform 9 is NFPA 130-compliant (thus, can clear a train in four minutes); as it is the narrowest platform, it follows that all platforms can clear a single train in four minutes, and (except platform 10) are only noncompliant because they can’t clear two at once. There’s no need for the silly 18-22 minute dwells that the LIRR and New Jersey Transit think they need. With more reasonable dwell times, NJT-PSA through-running (maybe also with some LIRR lines if by the timeframe we’re talking about – end of 2020s/beginning of 2030s – they can get rolling stock that can use both third rail and 25 Hz catenary, or else reelectrify Penn Station-Kearny with 60 Hz) takes two platforms and four tracks, Amtrak gets one platform and two tracks, the 33rd Street LIRR terminates on three or four platforms, and the station can thus lose the southernmost four or five platforms all at once.
Ad Grand Central: the Lower Level has room. The Lower Level tracks terminate midway between 43rd and 44th Streets; there are about 110 m from there to 42nd. The building west of Park between 41st and 42nd looks like 23 stories * 30 m * 60 m = 41,400 m^2 = $250 million to replace, which is bad (and unlikely) but not the end of the world, and by the time the tunnel would hit any other building it would be deep enough not to risk anything – and deep construction under private property is pretty common for deep bored subways.
Train-to-the-Game was scheduled a for ten minute stop in Penn Station. They really should check their own schedules. The northbound Vermonter, train number 56, for dates in October is scheduled for twelve. I stopped looking after finding that.
The very old, at this point, track map I have of Grand Central says there are only four tracks 13 cars long, on the lower level. The rest of them are 10 or less. The floor of the Lower Level’s concourse doesn’t sit on dirt. There’s lots of stuff lower than that. If it’s going to be under Park Ave why do you have to tear down buildings on Park?
The Silverliner Vs are supposedly capable of running on 25Hz or 60Hz. Converting to 60Hz has been on everybody’s wishlist for decades. It’s not going to happen all at once overnight. The NJTransit multilevel MUs will be. No need for third rail for things going to and from the New Haven line. Just like the trains that do it now.
I know exactly where MNR tracks terminate, I worked on ESA for 9 years and worked with MNR regarding the various interfaces including lengthening some of MNR’s tracks as well as dealing with the loops and all the other crap that sit under Grand Central. And there is a lot of stuff lower than than the lower level tracks that would be very difficult to remove. The tracks are simply not low enough to clear all of the stuff that’s in the way such as the subway which includes 7 Line, Lex Line, Shuttle etc.
As for Penn Station, you cant rebuild Penn Station without closing parts of it, I guessed at 50% just as a guess. Either way, there’s going to be disruption because whatever you say AMTRAK who control Penn Station are unlikely to change their attitude on dwell times.
The lower level tracks are pretty close to where the tunnel would need to go anyway, between the Shuttle and the 7 platforms.
And you don’t need to close 50% of Penn Station at any given time, only probably 25-33% at a time, which is fine. The “but Amtrak won’t allow it” line is an excuse – two tracks and one platform is enough for Amtrak’s current ops, and Amtrak at any rate is even more a ward of federal funding than the state commuter railroads. (It’s so silly how each railroad says “we can’t, the other commuter railroad won’t allow it” when they all get funding from the same sources.)
On the level of “secrets of Grand Central” type of short documentary videos, they usually cite “ten stories down” and then have a tour of the electrical hall were the AC is converted to DC for the third rail. I assume they mean under the headhouse, somewhere – most people don’t know the skyscrapers along Park Ave, in the 40s, are built over tracks. … the north concourse is under 47th? 48th? The building itself doesn’t sit on dirt there’s going to be foundation and it’s likely “down to bedrock” or “into bedrock”…. Most railfans think skyscrapers sit on street level dirt like their garage does. They don’t.
They want to move the Empire Connection. It was built through an almost abandoned railyard which was then rehabilitated to the LIRR West Side yards. And they’ve gone and built 20 billion dollars over skyscrapers over that. I seriously doubt that can be moved. I don’t know why it should be either.
Yeah. We extended Track 115 and had to put a new bumping post in. It is probably the most expensive bumping post ever built. Directly beneath Tack 115 is a canteen and beneath that is one of those service facilities, could be the converters, don’t remember exactly. There is about 60ft of structure beneath Track 115 and the minding post had to be founded in the base slab of that room!!! and there’s all kind of facilities south of the tracks which would have to be relocated/underpinned.
Its a shame that images don’t seem to be able to inserted here, because I have the long section through GCT and everything south of 42nd St that shows the foundations of GCT and all the stuff that’s in the way, because clearly there is a disconnect here. And there’s one thing I completely forgot about and that’s Burma Road, that’s a cross passage beneath the lower level that includes ConEd high pressure steam lines for many of the overbuild structures as well as multiple traction power cables for the lower level tracks, including the east storage yard area. In short there is no quick and easy way to go south from GCT, in fact I’m not sure its even possible to thread between all of the existing structures and facilities.
If you email me I can upload them and post links. (I should follow up on this post anyway, since somehow in this 3.4k word post I didn’t talk about phasing except in comments.)
here you go
I’m not seeing the email in any of my inboxes – which address did you send it to? (I guess if it’s a bunch of images it’ll take some time to show up…)
Huge fan of having the simplest stations possible, and absolute hate when stations are turned into giant shopping malls. I loved the trains in Japan, but navigating the stations was very annoying; it often felt that the stations were made artificially large solely for the sake of building more commercial space you’d have to walk through.
Some British airports are exactly the same – to get to your gate you have to navigate through racks of retail merchandise where they try to obscure as much as possible which way is “out” of the shop towards your gate.
Which is particularly fun when the insanely long security queues mean you only have a couple of minutes to get to your departure gate before they close the doors and you miss the flight you booked on a non-flexible ticket (the only kind of tickets they sell these days).
But airlines can only get away with all this hostility to their own passengers when there is no viable alternative to the connections they offer. As soon as a train line comes even close to competing with them on time, then patronage falls off a cliff.
Penn Station’s current users believe they need 18-22 minutes to turn.
If the trains are running through they don’t have to turn. The Train to the Game service had a ten minute dwell at Penn Station, Metro North would turn over the train to NJTransit or vice versa. Rumors on railroad.net are that NJtransit turns over trains going to Sunnyside, to Amtrak or vice versa in four minutes.
I just realized I didn’t see any restrooms. As a parent I can be confident that my kids will demand a bathroom now, and I expect to find one in the station. I’m not there yet, but I also know several old people with “bladder issues” who need plenty of restrooms scattered around.
Some trains will have them on board, but will all of them. With fast turns as a goal, if we are close to the station (which is likely!) I’ll make the kid wait until we get off rather than risk the train pulling away while I’m onboard. And of course for infrequent trains, or those with assigned seating: I’m going to arrive early so we will need restrooms while waiting for the train.
Railfans rarely consider that passengers and staff bring their stomachs, bladders and bowels with them. Some of them don’t even consider people and obsess about moving more trains instead of moving people.
Also for the concessions selling beer, if your headed to NJ they will have to keep a stack of brown paper bags handy, and cardboard cups for the wine drinkers, and for LIRR trains a covered vessel, just think of the potential for a mistake, arrested in NJ for not hiding you beer in a brown paper bag cos the vendor mixed up your order…… the shame….
What’s a brown paper bag? A curious German wants to know.
Certain states/municipalities have laws prohibiting open containers of alcohol in public. Exceptions are made for special cases (i.e. presumably inside a train ) if the open container is wrapped in a brown paper bag and thus consumed (“wino style”) so as not to offend sensitive souls.
I think Alon knows this. In Germany drinking alcohol in public is an enshrined human right, alongside bringing your dog to a restaurant and letting off fireworks on New Year’s Eve. They are thinking about bringing back the death penalty for crossing the road on a red light, however.
(The part about fireworks is changing now that right-wing media discovered that some people of color partake; the attitude changed quickly from “the Cancel Culture leftists are fighting our traditions for their silly quality of life crusades” to “criminals with migration background are disturbing our sleep.” A CDU MdB even started ranting about “phenotype” in January.)
Railfans rarely consider that passengers and staff bring their stomachs, bladders and bowels with them. Some of them don’t even consider people and obsess about moving more trains instead of moving people.
The acquisition and operating costs of sanisettes are, by the standards of what we’re talking about here, not even a rounding error. There’s plenty of room for them on the streets, on the walkways, inside Moynihan (where the gender-neutral bathroom was closed when I was there, to my great consternation), and even on the platforms below the escalators.
It is a rounding error only so long as you design them in, in the first place. If you don’t have the pipes (or have them but they are too small for the load) it can mean a somewhat expensive remodel to dig them all in. The cost isn’t just dollars: the cost can be reduced passenger flow, or just architecture costs as the plumbing/ventilation messes with the clean looks.
There is a whole science around restrooms (and in my experience it isn’t well developed).
Yeah, I think the bathrooms should be in the model, otherwise people will assume the station doesn’t include them and politicians might decide to build the station without them because something something homeless people and drug users. The NY Subway actually has old bathrooms that were closed, presumably because of fears of “crime”
I also realize I didn’t notice any places to sit in the model. New York already has too many places where they removed benches to deter the homeless, public seating in train stations is important.
Are you committed to the open-air concept (i.e. no roof at all for most of the station), or does the design leave the possibility open for a roof. Given that Berlin Hbf is one of the inspirations, I think a vaulted glass roof similar to that station would be a good addition: it retains sunlight and a sense of spaciousness, while also protecting from the elements.
And intangibly it helps to give the site the feel of a railway station, rather than just being a big hole in the ground. I know people on here are justifiably skeptical of architectural statements, but it doesn’t have to be a Calatrava-style monstrosity, just a functional glass roof, elevated enough to avoid making the platforms feel cramped.
Agreed. Nobody wants to stand in the February freezing rain to wait for their train, and there’s no reason why they should need to.
Rain drains away. The five-year snows that are knee deep will be interesting. The 10 or 20 year ones that are deeper, even more interesting. Standing around in the sun when it’s body temperature at the Central Park weather station would be good too.
I think–assumed–the whole “hole in the ground” concept was inclusive of some kind of roof. Given rain/snow’s propensity to much up both passenger mobility and (in heavy snows) train mobility in a place with NYC’s weather, it seems far too obvious to not include. I thought the idea was simply not to have massive initial build-out of facilities like the recent back-office bloat and mezzanines of other projects in the city. “Simple and functional” would include a roof in any modern high-capacity transit station.
There’s a possibility for a roof, yes. There are a couple different options there:
1. Three glass sheets (for the trisection of the main station by the two walkways), either at or somewhat below street level.
2. A high roof, turning the station into a recognizable single building.
3. Low shelter above the parts of the platforms that are exposed to the elements.
Well, I would definitely opt for option 2, particularly because it would also provide shelter for people changing between platforms, and to minimise the possibility of passengers feeling cramped on the platforms (which even a glass roof can produce if it is low-hanging).
I can also imagine that any plan that left the tracks exposed to the elements would meet with operational problems during icy or stormy weather. Granted, these would happen anyway on the outer lines, but Penn Station would be the most sensitive part of the network where reliability is at premium.
There’s a good example of this in Crossrail/Elizabeth Line at one of the central stations (can’t remember which but either Paddington, Bond St or Tottenham Court Road). They purchased and demolished a building to allow direct natural light to penetrate all the way to the bottom of these deep tunnels. It has a fancy hi-tech glass roof that is a bit of an artwork (vaguely recall it might be by Nicholas Grimshaw who designed the Waterloo terminus for Eurostar). The escalators are equal to the longest in London.
This was featured in some detail in the rather good tv documentary on building Crossrail. Oddly, couldn’t find it on the Wikis on Crossrail or the various stations.
Also, I would probably lean to going with 6 platforms rather than 8. Your 3-line model should be able to handle commuter demand for a long time to come, and with a bit of tweaking could probably accommodate up to 12 intercity trains per hour in both directions, which is a huge amount of capacity for intercity travel.
If there really does end up being demand for more than 12 x 16-car trains on the northeast corridor in several decades, then the railroad would be so awash with money that building extra platforms at Penn South wouldn’t be an issue.
Having 6 platforms/12 tracks initially gives you much more comfortable platform width and from a quick look at the current Penn layout looks like it would be easier to implement, since more tracks could be kept in situ (with some tracks being filled in to create wider platforms).
The reasons I went with eight and not six platforms:
1. The cost of Penn Expansion is ridiculous and getting even more so, so being able to safeguard against that is good. Much of it is inherent to the project – it’s a Manhattan block and around 300,000 m^3 of excavation – so cost control is possible but will not make this affordable.
2. Eight platforms are already wide enough for most things; the extra circulation space with six wider platforms helps little when the limiting factor is the circulation on the two walkways.
3. Both eight and six platforms let intercities overtake regionals if they need to, but it’s easier with eight.
4. Six platforms actually have a problem tapering to the east while respecting Penn 11’s footprint; with eight it’s possible with a 300 m curve radius and a bit of undermining, to the same extent as the existing taper toward 32nd Street.
1. Right, but when would this be needed? When would three bi-directional lines running through Penn not be able to handle capacity (in addition to the non-Penn lines you have sketched in maps)? 50 or 100 years from now? Is safeguarding for that far in the future really worth it?
2. If walkway circulation is the limiting factor, couldn’t you just widen the walkways?
3. OK
4. I’d have to take your word for this. But perhaps you could add a bird’s eye view of the station, overlaid with what is currently there, to see how much track-slewing, etc. would be required for 8 platforms.
In case you didn’t see this:
Penn Station expansion could balloon beyond single block, hit whopping $16.7B, new plans reveal
By Nolan Hicks
Published Sep. 19, 2023, 5:57 p.m. ET
https://nypost.com/2023/09/19/penn-station-expansion-could-balloon-beyond-single-block-hit-whopping-16-7b-new-plans-reveal/
these may be thought of as 7⅓th and 7⅔th
Why do I want to get to 233 West 33rd? Or 266? Way way back when the taxi stand was open, the only time I used it was when I wanted a taxi, which is vaguely at 7 and a 1/3. They allowed pedestrians to use it, almost no one did because it goes to 230ish West 33rd. Part of the ARC plan was entrances NORTH of 34th St. There is pedestrian congestion on 7th Ave now, it’s not going to get better if there are more passengers. There will be more passengers someday.
I can’t find the entrances to the 8th Ave lines. You are obsessing about getting the street, they should be at the subway level and 33rd or 31st St ends have access to the street. which is okay because more people want the subway than 230 West 31st. And it’s a good thing that people who want the subway are circulating on the subway, not the street. Because there is pedestrian congestion on 7th Avenue now. Or was before the pandemic. Probably still is because it was moderately congested 40 years ago when there was less ridership.
Your plan doesn’t show the subway entrance at 32nd and 7th. The closed passageway to the Herald Square subway complex was part of the 33rd Street and 7th subway entrance. Last I heard, part of the compromises with Vornado, to tear down the Hotel Pennsylvania ( east side of 7th Ave between 32nd and 33rd Streets ) was to restore that connection. Along with being ancient enough to remember the taxi stand I’m ancient enough to remember the passageway to Herald Square. It was quite handy. You aren’t in the hordes attempting to get to the street. And out of the weather. A concourse where 32nd Street should be would be good…. because there is a entrance to the 7th Ave lines at 32nd….
Why do the platform level signs say “JFK via Sunnyside Junction”. WTF does Sunnyside have to do with getting to Jamaica and JFK? If some yokel from the hinterlands asked me where the blue train to the airport was…. the first question I’d ask is “which airport” because there are trains to Newark Airport. And when I determined it was JFK I’d say “any train that STOPS in Jamaica, go UPSTAIRS to the Airtrain” Because they might want to go downstairs to those blue trains. …. why does anyone, who is going to JFK care that the train passes through Sunnyside, Woodside, Forest Hills or Kew Gardens? They care that it stops in Jamaica where they change to Airtrain. Upstairs.
Yeah, the street-subway interface is not part of the level – sorry about this. (To put things in perspective, the version of the level 12 hours before release did not have the subway entrances from East and West End Corridors.)
Sunnyside Junction is specifically the “via” for trains to Metro-North, since they can through-run either via there (from Platform 3) or via Grand Central (from Platform 1).
Trains going to the New Haven Line wouldn’t be my choice if I wanted to JFK. The only reason I can conceive of, to get on a New Haven Line train, is to get to someplace along the New Haven Line. In what bizzarro scenario do you imagine someone would take a New Haven line train from Penn Station to Sunnyside to change to a train that left Penn Station moments later to get to anyplace? Or go to Grand Central to clamber between platforms to get to Jamaica. If I wanted to get to Jamaica I’d take a train, there will be one in a few minutes, that is going to Jamaica. 24 hours a day every day. And even in the dead of night, a few times an hour.
Getting to Grand Central is great for getting to Grand Central or it’s very close vicinity. Everywhere else there are other alternatives, most of them “better”.
I can’t imagine why, except in some fevered foamer’s nightmare, there would be trains from Penn Station to Grand Central to the 63rd Street tunnel.
I would avoid having constructions below the escalators/stairs. The part of the platform which is narrower because of the elevators is typically a bottleneck and feels unsafe. Having retail there causes even more contention.
Having just the escalator/stairs would minimise this area. Having the stairs and the escalator at separate points in the platform removes this problem and also allows to better distribute the flow of outbound passengers better. What I would suggest is to have “bridge” extend a bit over the platform (perpendicular to the current bridges), with the escalators going down first, and then the staircases in a central position. That bridge can host cafés and similar retail, there is less contention and the view is better.
The narrower parts of the platform are still rather wide – it’s 2.8 m from the edge of each escalator to the edge of the platform, which is fine for obstructions of this length. Generally the escalators and stairs go together – it’s pretty standard at stations here, so that the wayfinding is easier and fewer access points are needed.
2.8 meters is not that much if you have a food shop with a bit of affluence, a clump of confused travellers with their mountain of luggage and someone with a double-width pram. For what I observed, people are way less willing to switch sides when one is very busy when they can’t see that the other side goes where they want.
The underground tracks in Zürich, in particular for Bahnhof Löwenstrasse (tracks 31,32, 33 & 34) use this staggered approach (you can see it clearly in the Halle Sihlpost) and IMHO this separates the flows of people nicely.