Radial Metro Design on Rivers
The most common and most useful design paradigm for an urban metro system is radial. Subway lines should be running across the city, passing through city center with transfers to other radial lines; larger cities can also support a circumferential line, or for the largest megacities (like Moscow) two, and unless there are multiple circumferentials, every pair of lines should intersect with a transfer. For example, here is Prague:
There are three lines, meeting in a Soviet triangle, running from one side of city center to the other. Together with an intact tramway network, this boosts Prague’s annual urban rail ridership to around 830 million a year, which is 310/capita, a figure that isn’t far lower than Tokyo’s and is higher than anywhere else I can think of.
But in some cities – but not Prague – there’s a kink in the radial design. For example, here’s Kyiv, with planned expansion:
The three existing lines form a perfect Soviet triangle. Line 4, Podilsko-Vyhurivska, is under construction and radial as well. And then there is the under-construction eastern extension of Line 3, Syretsko-Pecherska, looping back to meet Line 1 at Darnytsia. This is not standard radial design. But it’s fully understandable given the situation of Kyiv.
Kyiv has a division into left-bank and right-bank Kyiv. The Dnipro is, with islets included, 1-2 km wide, one of the widest rivers of Europe. There are few bridges. The main of the city is on the right bank, but left-bank Kyiv has its own independent center around Darnytsia, encouraged by the city’s development plan precisely because the river is such an obstacle.
The river division is not universal. Prague doesn’t quite have it – the Vltava is 160-200 m wide and there are many bridge crossings, so even though city center grew along the right bank, much of the near-center is on the left bank. The city is also hilly enough that there’s no coherent left- vs. right-bank identity, and the streetcar system is sufficient to connect left-bank neighborhoods with each other without passing through city center.
Conversely, London does have this division. Bank terms are not used there – one says North and South London – but the situation is the same, even though the Thames at 250 meters is not much wider than the Vltava, and has many crossings as well. Nonetheless, a South London identity exists, defined by paucity of river crossings to East London (but not to Central or West London), and by its own centers at Waterloo and London Bridge.
As a result, the radial Underground network forms a coherent sub-network in South London. Just as the Kyiv Metro is planned to feature a loop back on Line 3 in left-bank Kyiv starting 2023, London built the Victoria line to swerve east to cross each trunk of the Northern line twice, once in North London and once in South London, and the crossing with the main line at Stockwell is even cross-platform. Unfortunately, the South London crossing with the Battersea extension is without a transfer, a deliberate design decision made to reduce ridership and perhaps reduce crowding on the Vic.
Finally, New York should think explicitly in terms of right- and wrong-side parts of the city, the right side referring to city center, that is Manhattan. New York’s subway network is not radial, but the same principles apply just the same. There is a strong wrong-side identity for Brooklyn, and historically Downtown Brooklyn was a very large business center; today it remains near-tied with Long Island City for largest job center in the region outside Manhattan. Early-20th century designers did not think in such comparative terms but they understood that it was valuable to connect Brooklyn homes with Brooklyn jobs, and thus most subway lines in Brooklyn converge on Downtown Brooklyn, and only the J/M/Z and the L go directly from Williamsburg to Manhattan.
By a fluke, all four subway lines in Queens connect to Manhattan via Long Island City, the nearest neighborhood to Midtown. Thus, a business center emerged there, growing to rival Downtown Brooklyn; just as the city’s geography can create a subway network, the subway network can create the city’s geography.
Pedestrian Observations 
In Kiev, I think the current plan is worse than having the Red Line branch at Darnytsia and meet the Green Line at its terminus.
Advantages of current plan: the current Green Line tail gets a one-seat ride to Darnytsia
Advantages of branching the Red Line: the newly constructed segment gets a one-seat ride to the city center.
The city center appears to be much more dominant than Darnytsia, so branching the Red Line is a better idea.
So even in this case, the heuristic of “make your lines straight” is true.
(Also, branching the Red Line reduces frequency on both tails by 50%. But this is probably a good thing given the likely ridership on the tails versus the core, the operational savings, and the unlikelihood of the core ever surpassing its frequency capacity.)
Also, tram line 3 is basically all grade-separated already and would be a cheap and valuable conversion into another Red Line branch on the other side of the city.
Yes, and also ridership is high enough that frequency is high even if you cut it in half. The Soviet way of building metros is hostile to branching, and this is plausibly a place where it would have been good.
You must mean the Soviet culture. The geometry of Soviet triangles is perfectly compatible with branching – for example Munich’s metro is a Soviet triangle with all three lines branches (and the branches crossing each other in some places).
The geometry is perfectly compatible with branches, but the service planning isn’t. Timekeeping is based on pure headway management, with time-since-last-train clocks (Moscow’s famous for those, but I saw them in Prague too). Branching introduces a slight snag in that system; it’s fully compatible with spamming very high frequency, as on M13 in Paris, but it is a change.
Any thoughts on how this analysis would apply to the Bay Area, where San Francisco Bay is an enormous obstacle cutting the region not just in half but into several centers and corridors? Or is the area so large and divided that any attempt at Bay Area unified transit* below the regional rail level a fool’s errand? Is the Oakland-SF traffic so large that it is worth it to pursue an integrated subway network for them, and if so how would the 7km “river” between them affect network design?
*unified as in a single system/mode serving the whole Bay. An integrated system (common fare structure and scheduling so riding BART from Richmond to Berkeley costs the same as riding AC Transit, or so BART and Caltrain have timed transfers at Millbrae – aka verkehrsbund) should be pursued regardless.
Both SF and Oakland have (if you count Muni) radial networks which are integrated with each other. So no more work in this respect is needed.
BART’s issue is lack of transbay capacity, which means another tunnel between SF-Oakland is needed. This is easy to fit into the network in a logical way: see https://imgur.com/ggFrZFV
…this absolutely applies to the Bay Area. Good catch. The Bay is so wide that there is a definite wrong-bank identity called the East Bay, with its own center in Downtown Oakland. BART was designed with this in mind, just as much of the New York City Subway was designed with Downtown Brooklyn in mind. The pattern of the Richmond-Fremont(-Berryessa) line exists to ensure people in East Oakland have a direct train to their own city center.
Unfortunately, this design has a serious seam: East Oakland does not have a direct line to San Francisco via Downtown Oakland, and Berkeley does, which inverts the importance of Downtown Oakland to those two areas. There’s an article about this from the 1970s in (I think) The Nation criticizing this pattern, pointing out that 2/3 of Berkeley commuters worked in San Francisco and not Oakland, and yet most trains based on that era’s operating pattern didn’t take them to San Francisco; today, the trains are 50-50, and Berkeley commuters split as 12,600 to San Francisco, 7,000 to Oakland (3,300 to Downtown Oakland), and my attempt at drawing an East Oakland blob on OnTheMap yielded 15,000 commuters to Downtown Oakland and 30,000 to San Francisco, a much more SF-focused geography than was the case 50 years ago
So the issue in the Bay Area is not just about network design, but also job geography. The American pattern is one of relative decline of secondary urban job centers, despite their good access by mass transit. The challenge, because Downtown Oakland is so well-served by mass transit, is to encourage job growth right near those BART stops – to make sure Downtown Oakland behaves more like Long Island City or the Jersey City waterfront or Kendall Square (P.S. Cambridge, too, is a very strong wrong-side identity) than like Downtown Brooklyn or Downtown Newark.
As far as network design goes, any second tube design must serve Downtown Oakland, rather than cutting it off for some kind of direct Berkeley-SF service. But all the possibilities I’ve seen for the second tube alignment already do that, because Downtown Oakland is still a distinguished node and the physical geography of the Bay is such that it’s not even that easy to cut it off.
Downtown Oakland behaves more like Long Island City or the Jersey City waterfront or Kendall Square (P.S. Cambridge, too, is a very strong wrong-side identity) than like Downtown Brooklyn or Downtown Newark.
Downtown Newark or Brooklyn are Pre Revolutionary centers that had their population explosions before World War II. Oakland is at the center of Post World War II growth. Smaller CSAs are going to have smaller centers too. The San Francisco-Oakland-Berkeley MSA has less people in it than Brooklyn and Queens. Clump pieces of Northern New Jersey one way you get another one. Or smaller bits, the San Jose MSA. …there are more people in Brooklyn than there are in the San Jose MSA.
One geographical aspect is that San Jose now connects to the Peninsula (Caltrain) and to the East Bay (BART), and it doesn’t have the “river” problem. Could it become a location for a “Russian triangle”? The fundamental problem is that BART, Caltrain, and VTA had their ridership fall off a cliff and the ridership doesn’t seem to be returning. BART doesn’t need another tube now.
The Bay Area is very polycentric. By employment density pre-pandemic, downtown SF was the primary job center, with Oakland and Berkeley as secondary job centers. SJ / Silicon Valley would have been the secondary job center based on the raw numbers, but it developed during the era of cars, and thus its office parks and housing sprawl make it difficult to serve with rapid transit. Nonetheless, South Bay is still the second anchor of the region, which has benefited development along the peninsula cities between SF and SJ.
Based on these numbers, we can see that:
– There should be an urban rail line running between Berkeley, Oakland, and SF, aka the Richmond – Millbrae/SFO BART line, which is the most intensively used line and the only one with significant all-day ridership.
– There should be a regional rail line between SF and SJ, aka Caltrain. There are a lot of medium-sized downtowns that can drive ridership, even with WFH. Caltrain has high ridership potential once electrification is complete and trip times decrease, assuming a proper timetable that runs more local and limited trains. The Baby Bullet is less needed when SF employment and ridership isn’t likely to recover as fast as that at peninsula stations as South SF, at least until a DTX into Downtown SF itself is complete.
– Concord / Walnut Creek is a tier-3 center, whose proximity to Berkeley, Oakland, and SF justifies a regional line line, aka the Antioch – SFO/Millbrae line, which has solid ridership during commute hours.
– San Leandro / Hayward / Fremont (South Alameda County) are tier-4 centers, but their presence in Alameda Country (a BART district member) means that they received trains from the start. Unfortunately, the service plan (and resulting infrastructure) developed for them was one regional rail line to SF and the other to Downtown Oakland / Berkeley, when a single line to SF / Oakland (and a transfer to Berkeley) is all that was really needed. The rail line is also a nice stepping stone towards Silicon Valley, but an expensive tunnel under downtown SJ is not the optimal solution.
– Amador Valley (Dublin / Pleasanton) is a tier-4 center that normally wouldn’t justify a regional line, but it’s the gateway to the Central Valley and so it should be inevitably served (excluding BART) when a proper ACE corridor is established.
Some other thoughts:
– Operationally, BART is one radial line through SF with 4 East Bay branches. The infrastructure also supports an East Bay (Orange) line that acts more as a circumferential line. Like the G, the Orange line doesn’t have good ridership because most riders would take the other 4 radial services. Perhaps the most important purpose of the Orange Line is to enable frequent headways between Berkeley and Oakland/SF via the timed transfer to the Yellow Line at MacArthur, and to serve the Richmond and Fremont branches off-peak.
– On the other hand, Caltrain is anchored by SF and Silicon Valley / SJ, but doesn’t serve either well. At the north end, Caltrain doesn’t have a DTX to Downtown SF so it’s relegated to serving the secondary center at SOMA / Mission Bay, and (unreliable) Muni closes the existing gap. At the south end, Diridon Station is 1 mile from SJ, and Caltrain runs too much express service through the South Bay. That said, it does serve the busiest peninsula stations well, with the noticeable exception of San Mateo.
– For OP, the Bay Area isn’t dense enough to have urban rail, outside the major cities. SF already has Muni, Oakland / Berkeley had the Key System (replaced by BART and AC Transit), and postwar SJ could have built a successful VTA light rail system if it had chosen its corridors properly (e.g. El Camino Real, Stevens Creek).
Imagine a triangle or wedge of pie with the point pointing down. The point is San Jose. The upper left corner is San Francisco. The upper right corner is Oakland. This view makes the BART tunnel under the bay a circumferential link. Oddly enough this is link number three ( #3 !) physically (count based on road bridges). The first link, Dumbarton, has proposals that make sense and has a vocal set of fans. The sleeper in the set, San Mateo – Hayward (#2), could be the center of an east-west corridor from Modesto (Yosemite !) to Half Moon Bay (Pacific Ocean). The neat thing about #2 is that part of it is currently either a BART line (Dublin / Pleasanton) or a commuter rail service (Altamont Corridor Express) that is expanding north and south from Stockton (its eastern terminus).
In the above view San Jose is a hub served by four radials : Caltrain to San Francisco; BART to Oakland / Richmond (SJ outskirts now, downtown future); Capitol Corridor to Sacramento; and ACE to Stockton. A fifth radial, south to Gilroy / Salinas is being planned and a bus link in Salinas is in place. It may start running in 2025. (As always, Union Pacific may prove to be a stumbling block.) TAMC (Trans. Agency for Monterey Co.) is the lead agency. It could eventually extend to Monterey.
San Francisco isn’t that strong of a center, but it is still the central city of the SF Bay Area. Center doesn’t necessarily mean geographic center. An easy example is any city built on a ocean/lake such as Toronto probably has its center on the water or not that far inland, while the metro area spreads further inland as water is harder to build on.
The BART tunnel and Caltrain is mostly radial. Oakland to San Jose is mostly circumferential. This forms an offset radial/circumferential loop line. https://pedestrianobservations.com/2017/05/13/the-yamanote-line-a-ring-or-a-radial/
Alon used the Yamanote Line in that article, which is a much smaller loop than what happens in the SF Bay Area. However, the much weaker Tokyo Mega Loop is also offset for similar geographic reason: Keiyo Line is radial out of Tokyo Station before meeting up with the circumferential Musashino and Nambu Line, before turning back in on the radial Yokosuka Line back to Tokyo Station.
A loop this big doesn’t really make much sense to be thought of as a loop line though, which is why both your proposed SF Bay Area loop and Tokyo Mega Loop are both generally understood to be a collection of lines that you happen to be able to take in a loop.
The point of my comment was that only using San Francisco-centric models in your planning blinds you to opportunities elsewhere in the S.F. Bay area. BART has been a funding pig for years and has drowned out various projects that can’t shout as loudly. Their desire to ring the bay threatens Caltrain’s funding.
The proposed BART extension to Diridon makes sense because that’s a hub station. But continuing on to Santa Clara is unneeded duplication of Caltrain’s service. If more trains are needed between those two points then run the Monterey County train through Diridon to Santa Clara with the possibility of basing some of MontCo’s rigs at Caltrain’s yard. That solution is cheap (no new tracks !) and timely.
Keep in mind that San Jose’s airport is NOT growing due to a lack of available land, encroachment by San Jose (noise !), and is threatened by sealevel rise. An S.J. version of the OAC would be a total waste of time and money.
P.S. Our host has suggested merging BART and Caltrain. That would only work if the expansion planning were assigned to an independent agency (e.g. MTC or Calif. DOT aka CalTrans). BART’s mistakes – Colma, Milbrae, OAC, etc. – disqualifies them from a broader planning role.
Every thread gets hijacked by either BART or some word salad involving Queens New York, doesn’t it?
Anyway, the post is about metros, you know, the things that serve the inner zones of actual cities, with headways in minutes and crush loads most hours of the day. It’s not about cow towns and “hub stations” that barely manage to support infrequent bus lines, nor about regional rail or inter-city rail between low-density low-ridership nowheres, and it’s tiresome to read the same tiresomeness about spawlburbs in California all the time. I’m sure Alon will post something about that again some time, and we can all repeat ourselves then.
The problem for the Musashino-Nambu system isn’t the concept of a large loop, its that they don’t actually connect the peripheral centres except Kawasaki with the Nambu. Chiba, Omiya, Tachikawa, Yokahama etc. That’s partly a function of geography Tokyo itself is too strong to make major jobs centres in the inner suburbs. In particular the Musashino core of the surburban circumferal was designed as a frieght bypass and has some missed connections with several of the main regional rail systems. Still in the grander scheme these systems work fine, most circumferential trips are local hops between radials.
For the Bay area studying the Yokosuka line’s connections to Chiba would be more useful
Well the freight bypass was built to intentionally around where peoples are.
It was a calculated trade-off given the expense of building a railway through rapidly urbanising exburbia as it was. And given the Tokyo area’s geography I can’t say it was particularly bad one. There were stupider decisions made in those years (e.g. Namboku-Mita interlining).
Imagine a triangle or wedge of pie with the point pointing northwest. The point is Roxbury Township, NJ. The right corner is New York City. The left corner is Philadelphia, PA. This view makes the Amtrak Northeast Corridor a circumferential link. Oddly enough this is link number three ( #3 !) physically (count based on the New Jersey Turnpike highway and NJTransit rail line).
Okay, I’m being unfair, but I hope you see that you can make this argument for any arbitrarily chosen point some distance away from two strongly connected major regional centers. If you find yourself surprised at the number of existing “circumferential” connections, that’s a good sign you’ve picked the center wrong.
For your framing to make sense, San Jose would have to displace San Francisco as the local jobs hub. Google Downtown West notwithstanding, SJ does not have the available office space to do this, and it isn’t on track to do so either–SF added 50% more office space than SJ in 2021. And the South Bay already has a huge amount of low-density office space, particularly in the form of sprawling corporate campuses on the outskirts of San Jose, that represent major sunk investments and are going to act as headwinds against building a functional core.
To do transit-oriented development, you have to actually do the *development*, not just the transit. Nowhere in the Bay Area is friendly to the required densification, but if San Jose has to reserve 3000 parking spaces for every 7 million sq ft of office space it green-lights, TOD will remain a pipe dream. You cannot use the same land for both car storage and human use.
In any event, this discussion is all way too focused on long-haul commuter rail, not the kind of local mobility that you would need to permit people to actually live without cars (and thus enable a virtuous circle of densification). Diridon is still smack in the middle of single-family/low-rise suburban sprawl, and the transit link to the existing major employer campuses is weak. Like… it takes an hour riding light rail to get from Diridon to Moffett Park or to Infinite Loop. Forget regional rail; San Jose needs to build public transit that actually allows you to get to work (and to school, and to the doctor, and…) once you’re in San Jose. And the dense housing that will justify and take advantage of it.
Coming at it another way: trying to make Diridon a hub a la NYC Penn Station is a fine vision, but the scales at present are not compatible. Pulling in commuters from Gilroy–that’s roughly the distance from Penn Station to New Brunswick, NJ. Penn Station can pull that (and then some). But Penn Station has Penn Plaza, which offers 17 million square feet of office space *without crossing the street*, let alone the huge chunk of the rest of midtown Manhattan that’s a 5-10 minute walk. That’s the kind of density you need to support this size of catchment area. And that one NYC superblock represents a decade’s worth of office building in San Jose at 2021 rates–and the local political process does not seem disposed to allow the kind of development that would be needed to change that.
I get it, it’s hard, because everything has to happen at once. You need lots of destinations, which means you need dense building, but NIMBYs/BANANAs want to stay low-rise for ten generations; and without the transit, you have to build parking, which means you can’t get enough useful-space density to justify the transit. No one has a good solution for this. But you maybe need at least some kind of plan before you go redesigning the entire regional transit network based on the assumption that the problem is solved.
Anyway, all that aside, five different train operators on mutually isolated systems that happen to terminate mostly in one place does not constitute a well-designed radial system.
This view makes the Amtrak Northeast Corridor a circumferential link. Oddly enough this is link number three ( #3 !) physically (count based on the New Jersey Turnpike highway and NJTransit rail line).
NJTransit, SEPTA and Amtrak all use the the same tracks and stations. There’s no reason why a NJTransit train couldn’t go to Boston, turn around and go all the way to Washington D.C. Supposedly SEPTA Silverliner Vs could too. Over Thanksgiving weekend Amtrak uses MARC or NJTransit equipment to run extra trains between New York and Washington D.C. Someday soon CTDOT will be running electric trains from New London to Stamford. There’s no technical reason why they couldn’t run from South Station Boston to Grand Central. The only thing that’s stopping them from going to Penn Station is a few hundred yards/meters of third rail that would need to be installed in Queens. Or Amtrak upgrading from 25Hz to 60Hz. They’ve been running trains, using NJTransit equipment, from New Haven to Trenton, on Sunday game days at the Meadowlands/Giants Stadium, for years.
The New Jersey Turnpike doesn’t go to Philadelphia it goes from the George Washington Bridge, allllll the way up on 178th Street, in Manhattan to Delaware, just bit south of Wilmington. U.S. 1 and I-95 go through Philadelphia. It’s NOT an Interstate highway south of exit 6. Or even a U.S. highway.
https://en.wikipedia.org/wiki/New_Jersey_Turnpike#/map/0
Google Downtown West notwithstanding, SJ does not have the available office space to do this,
I’ll see your Google Downtown and raise you Hudson Yards. Or Apple Park and 111 Eighth, Google’s Manhattan office building.
https://en.wikipedia.org/wiki/111_Eighth_Avenue
Last I heard Google wants to gobble up the former Nabisco bakery. The NIMBY’s are up in arms that people may be working in a ……factory building…
Eh, the point was the copy change; sorry if I fudged the details a hair!
Actually, 111 8th Ave is a great example of what I’m talking about. In *one block* Google gets nearly 3x the usable square footage of its 42-acre Bay View campus. *That* is transit-oriented scale & density, and what San Jose needs to start pushing if it actually wants to become the jobs center/main destination of the Bay Area. But it won’t, because “parking” and “character” and “I’ve got mine, no one else can move here any more (and gimme the windfall from the suckers who try anyway).”
To a certain extent I get it with NIMBYs: more jobs means more stress on the already-unattainably-expensive housing market, and since most people need more space to live in than to work in–a desk is smaller than a bed–the housing supply never keeps up (not to mention all the poor-doors, inferior-units, etc. for the Chosen Few lottery-winners who get to stay, or the arbitrary means-testing cutoffs that intentionally cut out the middle class, etc.). Of course, my sympathy is substantially modulated by the way the same NIMBYs are resolutely opposed to building more housing, since they’re the prime beneficiaries of the resulting scarcity. But I’ve been gentrified out of neighborhoods before and yeah, sometimes you’d rather a place remain depressed and underdeveloped and available to you, than become shiny and modern for somebody else. So since public investment toward non-suburban (let alone public) housing has been a nonstarter in this country for fifty years, I guess we’ll all just clutch at whatever we can and yell at each other until the seas rise to take us.
Anyway. Google did buy Chelsea Market, but it’s unclear if they’ll be able to expand the building. But again, I think people are less worried about the space being used than about the salaries being earned by the people working there, which’ll only make the local market more eye-wateringly impossible. (As though Chelsea hasn’t been a lost cause since Bloomberg took office.)
Google’s Manhattan building wasn’t built in a vacuum. The suburban dream was somewhat different before cheap automobiles. The Ninth Avenue El started whisking people to suburban splendor in Harlem in 1879 and later the Bronx. Or the ferries to the suburban New Jersey railroads. Or even a cross town streetcar ride to the ferry to suburban splendor in Williamsburg. ….. to get that kind of density, in broad swaths, you have to build it before there are cheap automobiles.
I’m sure there are very very expensive places to park near there but almost no one uses them. It’s a one or two seat ride from anywhere on the subway or a two seat ride from anywhere on the Long Island Railroad or NJTransit. Plop that building down in San Jose it’s in the middle of an enormous parking dessert. And it’s unlikely there will ever be dense enough origins to support much mass transit.
The big mistake in the San Francisco Bay Area was tearing up the Key System in the East Bay. They should have kept the key system lines in Oakland-Berkeley in place like they did with the light rail lines in San Francisco. That way you would have a light rail system for the big East Bay population center and can use BART to get to a from the East Bay to SF.
I feel like this has probably been asked before, but why is it bad for circumferential lines to intersect?
I don’t think it’s bad per se, it’s just of little value. Basically, a system of radial and circumferential lines effectively forms a grid away from the CBD. Like any grid, you can get from anywhere to anywhere with a two-seat ride: one radial one circumferential. A second nearby circumferential line doesn’t add to that. And transfers from one circumferential line to another are not particularly valuable because the two lines are going almost the same direction.
Of course, this is a general heuristic – every city has a different geometry with different density nodes, few circumferential lines are exact circles, etc so in some circumstances intersecting circumferential lines might be justified. For example Moscow has intersecting circumferential lines (historical 14 and under-construction 11) and it’s doing fine.
Hong Kong’s network now, as it stance, is excessively bad in term of number of transfer required.
First, Hong Kong’s metro network start with two lines, the red line and the green line. Both are on Kowloon peninsula, north side of harbor. Because the harbor was wide, at the time of planning, there were zero road/rail connection, with all transportation of human and objects relied on ferry crossing. So there were separate center south of the harbor on the HK Island, and north of the harbor on Kowloon Peninsula. They figured the red line and the green line, which extend Northwestward and Eastward on the peninsula respectively, after merging at the center of Kowloon Peninsula, do not need the capacity of two lines to cross toward the Island. So they decided to only build tunnel for the one line, and the another line would need a transfer to reach Central, the CBD of Hong Kong Island and the city. In the meantime, the East Rail Line (light blue on current map), known as Kowloon-Canton Railways at the time, have been electrified, double-tracked, and have an intermodal terminal just north of the harbor, connecting to the city’s first harbor-crossing road tunnel. Thus passengers would need to transfer from rail to bus to get to CBD. Alternatively, since this commuter rail interface with the green line, passengers can transfer from the commuter rail to green line and then to red line to reach Central
Shortly after, Hong Kong opened Island Line, a line that run horizontally along the coast of Hong Kong Island along the harbor. It can be said as a radial line from the Central CBD too.
Then, into ~1990s, due to further development of the city, two new road and rail tunnel have opened in the city, one on the east side and another on the west side. The one on the east initially serve as a direct extension of the green line toward the Island line, strengthening the connection between east side of the city, but then converted into a purple line that extend into a new TKO residential area, and travelers between blue line and green line would need double connection too. And the line do not connect directly toward city center, but rather feed into the blue line running toward west, so all travelers through the tunnel to the city center will also need additional transfer. Meanwhile, on the west side, a new rail line in orange is constructed on newly reclamed land, connecting directly to the city center, with a branch serving the city’s new airport. In 2000s, a new commuter rail line to new residential area in the Northwest, named West Rail Line at the time, opened, but also requires additional connection through the orange line to connect to the Island, hence also sharing the traffic of two lines into one. The East Rail Line also have a feeder line named Ma On Shan line at the time, requiring connection to travel toward anywhere.
Hence, while the city’s system is famous for ease in transfer, it really require a lot of transfers between two random destinations, especially when involving harbor-crossing trips. Circumferential trips are also slow on rail, and is becoming a bigger problem as the city’s offices extend eastward and northward along blue, green, and red lines, as well as with multi million population now live along East Rail Line and West Rail Line and connection between the two rely on buses.
There are plans to improve connection by bridging the purple TKO line, as well as the orange line, through reclamed land on the northern side of the HK Island, in a project named North Island Line, which will also interface with the soon to be completed railway extension of East Rail Line toward the Island. However, the North Island Line project have been deprioritized, as the government now want to develop new CBD on the sea to the West of Central, as well as a new parallel connection on the very north edge of the city, paralleling Shenzhen’s city core with intention to develop another new business district there. But the planned line to the North will need to rely on West Rail Line or East Rail Line to connect toward Central, while the line connecting the Western CBD3 on the sea is expected to need connection to connect toward rest of the cities.
For the North island line, Hong Kong initially planned to use the “swap” scheme which would have swapped the eastern section of the Island line from Fortress hill station with the Tung Chung line extension and extended the TKO line westward from North point station to take over the western section of the Island line creating an almost soviet triangle between the Tsuen Wan line, TKO line, & Tung Chung line save for the lack of connection between the Tsuen Wan and Tung Chung lines on Hong Kong Island. This would have resolved a lot of the line connectivity issues and over loaded transfer stations.
Hong Kong later chose the less useful “interchange” scheme which continues the poor line connectivity, but preserves the east west continuity of the island line. Hong has had a history of swapping lines as they built out the network, I don’t know why they all of a sudden got cold feet on doing another swap.
1. The biggest official reason is it break off established ties within communities.
1.a. Communities facilities on the North shore of Hong Kong Island in, like hospitals, libraries, stadiums, schools, are constructed to the south of existing metro line, on the western part of the blue line, which are to be used by residents along eastern part of existing blue lines, compared to newly developed residential towns which have their own social facilities within their own development area and mainly visit town centers for business and commercial reasons which are mostly located to the north of the existing blue line. The swapping scheme will not just force most residents along eastern part of the line to change line at Quarry Bay station to reach these destinations, but residents of Fortress Hill station in particular will need to backtrack to North Point station to transfer. And the design of the two stations are also bad in term of such transfer. To transfer from Chai Wan to Tai Koo segment via Quarry Bay station to Tin Hau to Kennedy Town segment, there are three transfer corridors and each requires 4-7 minutes walk depends on where you start your trip. On the other hand, backtrack transfer at North Point station would not be particularly easy either since North Point station is currently structured to enable seamless transfer from purple line to Western part of blue line, which mean cross-platform transfer from platform 4 to 2 or from platform 1 to 3, but backtracking to Quarry Bay would mean transferring from platform 1 to 4 for outbound trip or 3 to 2 for inbound trip.
1.b. And since the north side of the line is now mostly things like office, which are more likely to be used by long distance commuters, which the purple and green line have larger and younger population share, than the smaller and aging eastern part of the city, it would also be kind of mismatched to send train on the North Island Line towards the eastern part of the island.
2. In addition, It will also break off the eastern side of existing blue line from rest of the old MTR network (red,blue,green lines, and the South Island Line). The three oldest lines and the new automated line are more useful to matured communities as contrary to the orange/purple/ex-KCR lines which connect more recently developed area, but North Island Line would not reach Admiralty or Central station and thus they would either need double transfer either via the new purple line or via the to-be-extended-east-rail-line before further transfer toward those other lines that form the old MTR network, or walk from Hong Kong station to Central station, which itself is ~5 minutes with conveyor belts, or walk from Tamar station to Admiralty station, which mean out-of-station connection through open footbridges which will also cost ~5 minutes walk. And that is in addition to the previous opening of the purple line which break off the single transfer from eastern part of blue line to green line. To residents on the Island, the orange line will become a circumferential line, striking pass the marginal area of both Hong Kong Island CBD and Kowloon area CBD, contrary to orange line’s main purpose for now to act as a radial line that connect residents along the line to Central CBD on Island, and Tsim Sha Tsui CBD through connection to the ex-West-Rail-Line.
3. Even bigger problem would be headway and capacity and physical spec.
3.a. The orange line, due to its nature of partial track sharing with Airport train, as well as physical limits on number of trains that can co-exists on the Tsing Ma bridge as well as in the Western Harbour Crossing Tunnel, and also power supply system, can only provide something like ~17 trains per hour. It not just mean residents on the east side of the existing blue line need to face additional transfer to most of their destination on top of reduced headway, but it also mean the entirely newly constructed North Island Line tunnel will only be used by ~17 trains per hours as opposed to ~30, which would have been the case if the tunnel is to be used by other metro line, hence the “swap” scheme effectively halved the new tunnel capacity it can carry, without any reduction in cost (in fact it even incur extra cost due to the needs of refitting existing tunnels/stations for the spec of the orange line)
3.b. The orange line was also designed as a higher speed line. Hence it have its accordance construction limit, and other specs. To allow their trains operating smoothly in existing blue line’s tunnel would mean closing them off for long time and paying some high amount of money for retrofitting. Which I don’t think the cost have been estimated yet.
4. On the other hand, the benefit of “swapping” scheme is passengers along orange line (and former West Rail Line) can reach eastern part of Hong Kong Island with one less transfer, and passengers of purple (and green) lines can reach Central area with one less transfer. However, in the “interchange” scheme, both of these transfer will be simple cross-platform transfer, which is much less inconvenient than all the aforementioned troubles, and that is in addition to the purple line itself extending to the northern part of the CBD will provide one-seat ride to the northern part of the CBD, thus better distributing passengers towards the new line. (As opposed to the “swapping” scheme, where passengers coming from purple lines are motivated to stay on the same train for one-seat ride, and passengers from eastern part of ex-blue-line will also be motivated to change train toward the new purple line due to their destination are mostly around the blue line)
——-
But from latest rumors, it seems like the North Island Line is not just delayed but also effectively killed. The proposed location of Tamar station is to be near PLA’s headquarter in Hong Kong, as well as government’s headquarter, legislative council, and site to future high court building. According to the rumor, the Hong Kong government is worried about the risk of “terrorism”, not just during operation but also during construction of such line, as well as its existence as a station mean people can easily enter and exit such sensitive area. The station’s premier might also extend underneath the corridor that connect PLA’s headquarter to its bayside military harbor and thus the construction of the line might negatively impact the Chinese military’s capability in the city, so the government is leaning against the construction of the Tamar station. And then, there is also the Causeway Bay North/Victoria Park station, which have already been studied when building the East Rail Line extension to Hong Kong Island, and was found it might requiring closing off part of Gloucester Road for up to 5 years according to study at the time, which is the island’s main artillery road. (The Island is not like other cities where there are multiple parallel roads, the only other surface road that run east-west around that area is the one used by tram and only have 3 car lanes each way, with 1 lane mainly used by roadside bus stations) Situation have improved since then with the opening of a new road tunnel that connect east and west side of the island, but the impact of potential long term partial closure of Gloucester Road still appears to be far too much for the government to accept. With the two stations proposed to be cancelled, the new North Island Line will only feature 1 station along its multi-kilometers new tunnel, aka the Exhibition station, which can hardly justify cost for the entire project with all the urban tunneling, with a single station’s prospective demand and ridership.
* Not “artillery road”, should be “arterial road”
Han River in Seoul plays a major role in its subway design and usage.
Line 1-5 serve the older CBD north of the River, while Line 2, 7-9, Suin-Bundang serve newer CBD south of the river.
Line 2 (busiest line) crosses the river twice to serve both CBD in a loop
Random question: do you think New York made a mistake by building a non-radial metro? While radial is usually the best form, the elongated shape, heavily gridded street layout, and wide rivers around Manhattan don’t really lend themselves to radial design, and the most of the lines outside manhattan seem to me to be radial towards either downtown or midtown.
The problem I have with radial designs is it forces long trips for people on the edges when they are trying to get to someplace geographically close by on the next line. Which is the point of circumferentials, but I’ve never lived in a city that had them.
People have a time (not distance, time!) budget for getting places, which is generally about half an hour. If you figure 5 minutes to walk to the station, 5 minutes for a transfer, and 5 minutes to walk where you are going, that leaves 15 minutes for the transit ride. Could be 8 minutes in and 7 out. Or you could add a second transfer, but now you get 10 minutes for the total transit time which across all 3 lines. (above times are to make the math easy, feel free to play with them, but no matter what they make the point: people shouldn’t spend much time on transit) Which is why density around those radial points is so important as it shortens the time spend on transit for anyone going there and thus they can live farther out and still make it to you.
A few weeks ago someone on reddit posted the fractal network, which I think is the right idea. However the complexity of the picture combined with how regular it was (compared to a city which is always messy by nature) meant that nobody got it. Most cities planning transit are far too big to send everyone in the suburbs to downtown and back out. Instead they need good transfers to local lines for people not going downtown, and those places of transfers should develop into their own mini center.
Every city has circumferentials. But generally they are buses rather than rail. Only the largest cities have sufficient demand to make them metro rail.
In general you need a radial network to the core, plus a grid or polar network covering the entire metro area to enable anywhere-to-anywhere trips. These don’t need to be the same mode though. The radial lines have higher demand and thus generally get a higher quality transit mode.
Des Moines IA doesn’t have circumferentials. Nor does the Twin Cities (which is large enough to need several)
I agree that radial needs higher capacity (while this implies high quality, quality has many parts: some parts of quality the low demand routes should have as well). However the service on lower demand routes is what really makes the system useful, the better you do at getting people to realize they can use transit the more likely they are to use it for everything, even those rate trips on low demand routes, so you can’t ignore them or give them too bad service.
Even Des Moines with its pitiful bus system has a couple circumferential routes and the Twin Cities have tons of them.
While a few lines run in a circle pattern, and there some non-raidal lines – I don’t see any thing that actually circles either city. Other than maybe downtown where it doesn’t save much time vs just taking a radial to the downtown station and transferring (though the area is probably dense enough to support the line – I’m not arguing get rid of it).
In any case my point is that the suburbs don’t have good transport between them in general and so you end up having to drive if you work in the suburbs, even if a bus goes by your house and our office, a 15 minute drive is 1.5 hours by transit. While suburbs as less dense places won’t ever get as good a service as dense areas, the way transit treats them means that you have to have a car to get anywhere in a reasonable amount of time.
A line doesn’t have to cover the whole 360 degrees to function as a circumferential route. It just has to run perpendicular to the radial routes in its area.
The problem with cirumferentials is that you need sufficient frequency on them to make it worthwile to use them—If I have a 10min interval to town, a change and another 10min interval back out, the ‘circle’ better has to run every 15min or better few people will bother to check the timetable and time their day around such an infrequent route if show-up-and-go frquency is available via the centre (unless the ‘circle’ saves a *significant* amount of time).
If the important ones of your radial routes are running only every 20min or more infrequent, then don’t bother with cirumferentials; you have bigger fish to fry.
There’s a “because” missing in front of “few people will bother …”
You need frequency for another reason: it is very hard to make all the timed transfer work unless at least one (probably radials as they have more demand). Without frequency you end up with a 5-10 minute trip to the next radial line, and then a long wait for the train to arrive.
It might seem like with very careful upfront design you can make it all work, but it becomes harder and harder the farther out you get – radials get farther and farther apart (unless they branch – which gets interesting in a different way), and so the travel time for inner cirumferentials is much less than the outer ones to get to the line. What makes it impossible is nobody has a new city with unlimited budget, over the years of building lines where you have money eventually some of todays plan’s won’t make sense in context of what the city is, but you can’t abandon them because then the plan doesn’t work.
Of course frequent lines is a good thing anyway, so hopefully you can look at this objection as pedantically true but not important in the real world.
only the J/M/Z and the L go directly from Williamsburg to Manhattan.
They had service into Downtown Brooklyn on the Myrtle Ave El which would go over the Brooklyn Bridge. That was cut back to Jay/Bridge with a free paper transfer and then bustituted.
The Williamsburg Bridge relieved the capacity problems the Els were having going over the Brooklyn Bridge. The Chamber Street extravaganza was going to replace the Park Row El terminal. Or the ferry, the line served the ferry terminal too.
https://en.wikipedia.org/wiki/Broadway_Ferry. Delancy/Essex and Bowery were substituting for the streetcar to/from the ferry.
The Manhattan Bridge relieved the capacity problems the Els were having going over the Brooklyn Bridge. they had service to Chambers too.
….And service to the ferry…
There have been fun service patterns where the Jamaica train, today’s J/Z goes through the Montague Street tunnel. I vaguely remember that rush hour Brighton trains would swoop through the Nassau Loop and go to Fourth Ave too.
…Pointing the IRT at the LIRR terminal in Brooklyn was to encourage people to change to the subway there instead of at Park Row. From the Els.
There is no definitive explanation of the “Second System” There are hints that the companion to today’s G train, where it could be the F train’s local, was going to replace the Myrtle Avenue El. There are also hints that they were considering 8 tracks on Sixth Avenue. The initial plans were to put the H&M, today’s PATH, under the four tracks of today’s Sixth Avenue lines. That would leave space for two more express tracks on the lower level. Finagle something at 145th? Squint at the enormous concourse between the 8th Ave and 6th Ave platforms at West 4th, someplace to put the Sixth Ave super express?
all four subway lines in Queens connect to Manhattan via Long Island City,
Define “line”. There’s the Astoria Line, the Flushing line, the Queens Blvd line, the Jamaica line, the Fulton/Rockaway line and the Myrtle Ave El just makes it into Queens.
There is life east of Jackson Heights. The A, C and J/Z go through East New York and the M train …………..goes over the Williamsburg Bridge along with the 53rd Street tunnel. Queens to Queens. Just barely but Queens to Queens. If they had the urge today’s E train could go down Sixth Avenue instead of Eighth Ave ( swap with today’s M ) and go back to Jamaica. They don’t want to because they’d have to run shorter trains.
Finally, New York should think explicitly in terms of right- and wrong-side parts of the city, the right side referring to city center, that is Manhattan.
Even though you’ve spent a lot of time there, you don’t comprehend the scale, do you? The goal in Manhattan is to keep people out of Midtown so that other people who want to go to Midtown can. Wall Street is the country’s third or fourth largest business district depending on who is measuring what. Son of East Side Access under Fulton Street that connects New Jersey to Queens means people are changing trains in Jamaica, Newark or Secaucus instead of Penn Station. Spend a lot of money they can be changing for Wall Street in Flushing too.
14th Street/Canarsie line trains to Secaucus gets them out of Midtown and Wall Street. Average daily ridership, using 2018 counts, for 14th/8ve, 14/7th&6th and Union Square, would make it the country’s 8th busiest system. Itty bitty obscure PATH is the 6th busiest. People who want to go to Grand Central don’t have to stop at Penn Station. Extend East Side Access to Secaucus , they can change for Grand Central. … the ridership projections for East Side Access are comparable to Atlanta’s MARTA, the whole system. Keep the scale in mind.
You use rivers, but this thinking seems as though it should apply doubly so to bays. There’s the above mentioned San Francisco, but other strong candidates for this design Philosophy include Rio de Janeiro; Hong Kong; Sydney; Auckland; Florianopolis, Brazil; and Mumbai if India ever wanted to try transbay development.
Given the cost of some of these longer crossings perhaps the simplest thing to do in a lot of cases is to do cross platform transfers between the first two lines built on each side? (One can easily imagine a San Francisco Bay Area, where this was the case based on two level Market Street Tunnel.)
Istanbuls trans-Marmaray Tunnel solution? Focus on high-capacity Regional rail?
> larger cities can also support a circumferential line, or for the largest megacities (like Moscow) two, and unless there are multiple circumferentials, every pair of lines should intersect with a transfer.
Speak of which, how to calculate what’s the optimal distance for a circumferential lines? I recall it was previously mentioned that some waterside cities like Tel Aviv do not have properly distanced subcenters to build a circumferential line around them, but how to tell what is the proper distance? Is 10km, 30km, or even 100km, all acceptable distances? does this depends on population number and how spread out a city is?
I don’t think there is a proper distance. For starters, you can rely on your circumferential bus routes at a normal bus grid spacing. If some of these get more traffic because they link subcenters, and this traffic level is enough to support rail, then you can convert the route to rail (like Maryland Purple Line). Generally I think the decision to build rail depends on the ridership of a particular corridor and is independent of the characteristics of the rest of the system. But to some extent, building a circumferential line outside the core can mildly decongest the lines within the core.
It all depends, I don’t think there’s a hard and fast rule. I’m squeamish about 30 km, and 100 km isn’t even really one metro area even in Tokyo or New York (the majority of commuters at that distance don’t go anywhere near the center, and subcenters are more important).
The main circles I know of are, with their circumferences:
Moscow Line 5: 19.4 km
London Circle line pre-Hammersmith extension: 20.8 km
Osaka Loop Line: 21.7 km
Beijing Subway Line 2: 23.1 km
Madrid Line 6: 23.5 km
Paris Lines 2+6, ex-tail: 24.5 km
Shanghai Metro Line 4: 33.7 km
Yamanote Line: 34.5 km
Berlin Ringbahn: 37.5 km
Singapore Circle MRT, inc. stage 6, ex-Dhoby Ghaut: 37.5 km
Seoul Subway Line 2 main ring: 47.7 km
Moscow Central Circle: 54 km
Beijing Subway Line 10: 57.1 km
Moscow Line 11, inc. under-construction parts: 58 km (I believe) ex-radius
15-ish from Times Square in Manhattan is vaguely the city line between the Bronx and Westchester, Flushing, Jamaica, Ozone Park/far western edge of JFK airport, vaguely Kings Highway in Brooklyn, St George, Newark, Paterson and Hackensack. Railfans may think there is stupendous amount of demand but there isn’t.
100-ish from Philadelphia? Allentown, New Brunswick, Atlantic City, Delaware Bay and lots of farmland in Delaware, Maryland and Pennsylvania, then Lancaster and Reading. ….nah, I don’t think so. Not when Allentown doesn’t even have train service.
15 km north of Times Square is not the city line. It’s Marble Hill. 15 km west is Jamaica or Flushing. Probably is good demand there for Rapid transit if not a whole circle.
“ish”. It’s 15 from Grand Central. I looked at a fifty mile radius map. Quaint paper thing that used to be quite common. Not precisely because I was eyeballing it not getting out a ruler.
100km from Tokyo is roughly the main cities of the three Kita-Kanto prefectures, Mito-Utsuonomiya-Maebashi-Takasaki.
The three cities, two with Shinkansen service to Tokyo and the other with nonstop conventional express trains, are all facing population outflow toward Tokyo as all the interesting things happen there.
And even for those continue to live in those cities, as the only rail line connecting them together are slow local lines, and that it is no faster to go to each other’s city center than to Tokyo, with Tokyo obviously also offering more choice in term of any things you might want from a city, residents there also have little demand to travel between these cities.
I think there were a few buses connecting them together before coronavirus, but that’s it.
Kita-Kanto’s where you see the problems of Japanese economic and urban planning, sprawl, weak universities, poor transit co-ordination even though proximity to Tokyo means they are doing relatively well compared to most Japanese prefectures. (Tsukuba is the exception that proves the rule). There is a lot of good work that would worked wonders if done in the high-growth period. Convert the Jomo line to tram-train, build enough passing loops on the Ryomo to permit expresses and through run the lines the lower frequency lines coming out of Takasaki, Mito and Utsunomiya in s-bahn fashion.
Isn’t Kita-Kanto actually underperforming compared to some of the more remote prefectures, due to Tokyo’s heavy straw effect?
I doubt rapid service on the lines would help when not even highway bus can sustain their service after opening of expressway.
And the national railway that connect the area together also underperform compared to private railway that send people there to Tokyo
That’s definitely true in some the statistics which don’t adjust for commuters. And “remote” is relative, Hokkaido is remote but Sapporo is relatively healthy. Yamanashi is closer to Tokyo than Tochigi but the latter does better.
Yeah too late to much now. That’s why I said the 1980’s when growth could of been shaped. The Ryomo corridor’s Kanto’s rustbelt, decaying textile towns whose one asset is Subaru’s based out of Eastern Gunma. Also shows the costs of rail segmentation, JR has the circumferential network but Tobu has the radial to Tokyo and they do not work together!.
They do work together, but only on express train to Nikko.
And 1980s being end of bubble era the population growth already slowed down, and JNR finance at the time was also in no shape to do any investment to improve service, in addition to the toll of motorization, mean even that was already too late. Any meaningful improvement there that can hope to change their fate need to be completed by 1960s, and that would mean planning have to start from before WWII. And at that time there were much more capacity and speed problem that can see improvement than focus on vitalizing rural area.
Taipei’s Circular Line will be 49.2km once all four segments are complete (in 2035 hopefully…?) This is pretty surprising to me since I don’t think of it as being particularly far out from the center. And like the Yamanote line its eastern portion if more radial than circumferential.
If you take the center as Taipei Main then it’s pretty far out. Even Xinyi is over 4km out, and Muzha, Xindian, and Neihu are even farther- up to 8km in Muzha’s case. Plus there are a several weird turns.
Taipei’s loop should work in theory because it’s almost all high density right up to the mountains, and the loop can fill in awkward gaps left by the radial lines. In reality… those transfers are terrible (what the hell were they thinking with Banqiao?), the sharp curves slow things down, and development has not been at all transit-oriented (sure is handy that there’s 2 metro lines going to Xinzhuang’s driving school!).
Another example is Liverpool, strong wrong side identity (“Wirral”), wrong-side city centre (“Birkenhead”). Only one mass transit line crosses the river (in tunnel) and that then branches to form the entire wrong-side transit network, so you have a radial network around the wrong-side city centre, which then debranches into a single tunnel to the right-side centre, where it has connections to the right-side network.
Hard to think of a better example than Liverpool to be honest
Both Manchester (Salford) and Newcastle (Gateshead) have strong wrong-side identities, but way too many crossings for the transit to be much affected by the river.
What is Minneapolis-St Paul like? I think of that as a big river.