RR503

I neither think that WMATA is a good example of this given how fraught DC governance is in general, nor do I think that that system is uniquely bad. Let's not forget that WMATA was a relatively well managed system by American standards until the 2009 accident. As for Williamsburg, I totally agree that in the long term we need more capacity/more housing in those areas, but it's equally important not to lose sight of just how far below capacity existing lines are. The runs 20x8 and the 21x8. If you move both to 30x10, you've nearly doubled your capacity. Getting the via Bridge would also help unlock potential along the Jamaica Ave section of the .

Here's how you make that work: https://www.sciencedirect.com/science/article/abs/pii/0967070X9500022I Why would you not just do 6th Ave express to Williamsburg Bridge? Keeps costs to only lolzy levels.

S/b s diverge at the top arrow, run along the track pointed at by the middle arrow (called B lead), and rejoin at the lower arrow.

Likely an equipment transfer to CIY for some sort of maintenance work.

You'd be able to curve over to 3rd if need be, though underpinning won't kill if you just go deep.

....or that! Wouldn't be surprised if the marginal load movement and time benefits were greater.

Beyond SAS 3 being little more than a pipe dream at this point, I fail to understand why the conclusion from these facts isn't just that we need to redesign SAS 3. It's terrible planning that'll worsen the reliability of the system while adding no capacity. It's connections as you point out are weak; the platform that goes the furthest east is Lex-53, and as you point out that barely gets near the . I'd argue they need to redo the whole thing, placing the route under 3rd north of 34-42, and under 2nd south of there, terminating at Lex-63 for now/in Bronx or in Queens in the future.

Your plans are good, I just wanted to push back here. My apologies for being harsh. The issue with your logic is twofold. Firstly, 36tph is a high throughput, one that is _definitely_ impossible without deinterlining, which is to say your interlining will destroy its enabling condition however limited it may be. Secondly, the existence of more tube capacity doesn't change the basic fact that you're just redistributing throughput inefficiently. As I feel I'm always saying, there are ways to fully utilize Queens crossings with current infrastructure -- appending more infrastructure throws off that balance, and reduces the efficiency of the system as a whole. We should be aiming to maximize that efficiency. If we can have a 4 track SAS, we sure as hell can have a new 2 track tunnel to Queens.

If we spend 10 billion (or whatever that'll cost) building a subway line, we should be making damn sure we get as much bang for our buck as is possible. SAS 3 adds *zero* new capacity to the system; it needs to be redesigned to make sense.

They did just that at BG. Unsure as to exactly how the GOs panned out, but it's doable. Absolutely. There's a good number of folks who do to [some other Brooklyn line]; if you can combine better SAS frequencies with deinterlining at 34 and, potentially, Dekalb and the inherently more efficient Manhattan Bridge route, you'd be able to make an impact.

There was a study of this sort done for Lex-53 in the ‘80s. I believe @Union Tpke has a copy. This is definitely something that should be looked into as Midtown East gets even bigger, as well as the 59-63 link because that’s two free platforms right there.

This is a pretty terrible idea. Let me say it again: a properly designed network with current infrastructure would fill _all_ capacity in existing Queens tunnels. The SAS/63 interlining is bad enough from an ops/capacity perspective. Adding more jury rigged connections into the rest of the tunnel system? Please, dear god, no. Queens needs more cross-river capacity. SAS 3 needs an outlet that doesn’t involve destructive interlining at its northern end. It’s a match made in heaven. (And this is, of course, before we consider the complete impracticality of meshing some sort of connection into tunnel infrastructure.)

No, you're totally right that that's the optimal solution coming south from Harlem/Bronx in the AM peak. It's just that it has knock-on effects when extended to the rest of the line -- in Brooklyn especially, as you point out. That said, the determinants of schedules are much more Rogers/142/149 than they are convenience. At Rogers, s have to be snaked between s and s; at 142, you have to make sure no conflicting moves are scheduled through the plant at the same time, and at 149, you have to take the result of the previous two interactions and make that 'mesh' with merging service. Those three variables are a challenge enough; I doubt all that much thought is given to load balancing after managing them in the peak hours thanks to their complexity.

I left out the because its volumes will exist independent of any deinterlining plan -- if anything, deinterlining will route more pax via local. Agree re: Woodhaven. Having 36 as either an express or as a Dekalb-style station with the on the outside ( would then relay to some tail tracks built beyond) would be nice, too, as it saves you a less-than-optimal diverging move at Queens Plaza and, if you go with the option, sets up a Northern Blvd subway.

To a degree, yes. When you make that chart with the and together, the chart changes, but that's precisely the issue -- the complex interactions the has with the and force it to run irregular headways to maintain a schedule that has any chance of being delivered, which force the to run irregular headways, etc. Because trains don't generally line up into perfect windows between other trains, you're forced to do ugly shit like this, or schedule delays. South of 96 St, the loading difference between and trains is relatively small, and there exists significant intra-segment ridership between 96 and Franklin. Scheduling the unevenly creates load imbalances and increases the chance of dwell congestion related delays as you get further away from 135 St. Outside of Rogers, I don't see scheduling as being nearly as much of a problem as some B div lines, but it's something worth keeping in mind when you try to estimate the impact of interlining on rider-experienced service. Preaching to the choir here. Merges suck, and a delay at one generally leads to delays at the rest. I will say that E180 isn't much of an issue in the AM -- s use B lead to get around s crossing to the middle -- but the rest of them are crap. Rogers especially. But the infrastructure of today is what we're stuck with, so the task at hand is coming up with ways to mitigate these design issues as best we can. New switches and deinterlining at Rogers are at the top of my list for this very reason.

There’s a good bit of room to up service levels to handle more folks at 51. The question is much more whether or not the IND platform can handle the additional volume.

There is in fact a _lot_ of excess capacity during peak hours on the subway. We run 373 trains out of 600 possible, assuming 30tph on each core-bound track, and the cordon track with the most peak-hour trains (Lex exp) runs a measly 26tph. And yes, interlining can absolutely move the needle on those corridors. Especially in the PM peak, QB express dies because of that merge at 36 St; remove it, add CBTC (as they're doing) and you have a whole new railroad, one that's likely capable of 34-36tph if you can learn to do better-than-terribly with dwells. Same goes for Lex and IRT 7th with the junctions at Rogers and, to a lesser extent, 149 -- Rogers is second to only Dekalb in its capacity loss, and that's a loss that carries largely up 7th Avenue onto trains that are actually now _more_ crowded than their equivalents on Lex. I'm a bit skeptical that deinterlining will screw Roosevelt. The dominant transfer flow there is the wall of people dumping local for express; attaching attractive destinations to the locals will lessen that impact. To be sure, you'll get folks from express stops wanting to transfer to locals, but I'd be somewhat surprised if that ends up eating up the savings. That said, I absolutely agree that deinterlining should be accompanied with investments in station infrastructure if only because there's a lot of relatively cheap rider minutes to be had in things like the addition of stairs/escalators.

Back-to-back trains are yet another consequence of interlining. If you're trying to thread service at uneven frequencies through complex arrangements of merges, you sure as hell are going to get some ugly gaps in the schedule. Compare and contrast: ...the 's scheduled headways at 125 St ...and the 's scheduled headways there:

Not for Lex express, for other Lower Manhattan routes, for example the Fulton-area xfers. Those absolutely aren't the most direct routes, but they're a) one seat rides, and b) generally don't experience the crowding/congestion that Lex does -- both of which shoot run and transfer times upwards. Optimally we'd have a functioning Lex express, but even with CBTC, that route will remain slow in some areas; certain idiosyncrasies in CBTC control logic may actually increase dwell delays in areas like Grand Central. This is a fair point, but that market really isn't Lex express's sweet spot. For destinations north of Union Square, you're better off taking the from 59/51, and south of there, you're bracketed with Queens-linked one seat alternatives except for at Astor Place. Given how messy the GC-Union Square section of Lex can be, I would highly doubt that you save any time relative to the other route alternatives by doing - - . As @Around the Horn pointed out, more service variants = longer waits, and more interlining = less capacity. The impacts wrought in those areas by interlining are really non-trivial, and will likely get worse under CBTC. There's always going to be a tradeoff here, and there's 100% a balance to be struck, you're right, but again, in capacitally/operationally stressed areas, deinterlining is a necessary part of the solution. Metrocard O/D data sure isn't perfect, but applications thereof have rarely been off by more than a few tenths of a percentage point esp when overlaid with granular data sets like Census LEHD or CTPP. Would it be better if we had 'real' OD data? Sure, but it isn't like we're flying blind either. And we absolutely do have more than annual cordon counts: traffic checkers are always out and about, we have 6-minute resolution AFC data. All of this is to say we absolutely can speak authoritatively on whether things are improvements or not, maybe with a lesser degree of accuracy here, but nonetheless a good, first-order level of accuracy (which, btw, cuts both ways -- your claims of the virtue of something are just as weak as my claims of its failings if you denigrate data). I disagree on the . For one, you can't turn at Whitehall unless you're cool with the Broadway local below (at least) Canal only getting 6-8tph; Whitehall ranks *high* on the list of shit terminals in this system. For another, as I covered here, it's exceedingly difficult to deinterline Broadway without also deinterlining Astoria, at least before the corridor gets CBTC. These arguments are all secondary to the fact that the reduces Manhattan-Queens capacity by its throughput + some value related to its impact on the services with which it merges. If you can get a functional terminal in Astoria (which is looking likely), trains are going to be taking up valuable slots in both 60th and 63 Sts -- 63 St because whatever runs via 60 from QB is something that isn't running via 63. Given the massive growth in LIC/Astoria/Queens generally, making sure our trans-river capacity is maximized is something of the utmost importance, and the stands pretty directly in the path of that happening. Now, of course, there are short(er) term capacity opportunities that can be leveraged to increase service levels -- terminal operations at Forest Hills being the most important one, capable of contributing up to 10tph to throughput -- but I for one am skeptical that we'd be able to realize those benefits with the complex Queens Plaza-area merge arrangement in place. That's, again, a significant capacity loss, and is again one that won't necessarily improve with CBTC. The impact of increased service on 60th St tunnel ops would also be somewhat significant. Because the enters the tunnel at 20 where normal trains are going...much faster, the merge impact is even greater than it would be normally thanks to control line length issues. So again, IDK how possible that'd be without deinterlining. The question then becomes one of just _how_ many people are using the Lex 59 connection, how large they're benefit is from said connection, and whether that outweighs the losses to the rest of the system because of it.

I've yet to be shown that that dynamic would be the same what with Lower Manhattan having declined so much relative to Midtown since the 1950s and alternate transfer having been built out. You'll have to evidence this. And again, 63-59... Gosh, could it be that...the express is faster but that incentives just don't line up? I would love to hear you explain just how a rider from Woodhaven to 5/53 would make that trip under a Queens deinterlining plan without, ya know, staying on the local. There's a big difference between people hyperbolically evidencing their estimation of the impact of timers with suggestions that local > express and people saying that the current route structure in Queens heavily incentivizes transfers at Roosevelt. Ah, the good ol' anti-professional planning argument. We love it. I can't speak for you, but I care about things like whether or not the subway network in Queens will have capacity to absorb further growth, whether or not we can even _schedule_ even headways, whether or not an incident on Broadway can bring the entire B division down, and whether or not dwell issues at Roosevelt make FH-Roosevelt runtimes on the E and F look like this (lines are 25, 50, 75 percentile): ...so do pardon me for suggesting some data-driven approaches to easing the subway congestion and capacity issues in Queens. Because what we have sure as hell is not working.

This mentality sorta _has_ to be broken for any real change to happen. Transfers are shit in NYC because service frequencies and reliability are shit, in part because we try way too hard to guarantee one seat rides. The negative impact of merges on speeds, capacity, reliability etc cannot be overstated. As for the Broadway-QBL issue, this is why you build a 63-59 passage and/or deinterline so that QB local goes to Lower Manhattan via 8th to get those Lex riders a good replacement service. Or you don't do those things, and understand that giving everybody everything they want is shitty planning -- 59 St is a serious problem area for Lex. Sometimes you have to cause a small(er) group of riders pain for systemwide benefit; taking a myopic view that focuses just on losers is exactly the mentality that got us into this mess in the first place.

Dunno about that. The data I'm looking at for the show a solid 2-2.5 min median PM peak runtime premium from Canal to Dekalb. When the *median* loss is that much, yeah, you're inconveniencing people. Given the interchangability of the 6th and Bway corridors, the fact that the only people who'd really lose full dual access are folks at Brighton express stops and 36 St, and that there may be people who want the corridor they don't have, I'd def say this is worth considering. Also the reasons @Around the Horn and @T to Dyre Avenue give. Capacity and schedule reliability are important!

That was a decision made pre-Bleecker St, pre-signal mods and pre-service performance data in an era when job distributions looked quite different than they do today. I would not be at all surprised if the rider mins equation favored deinterlining today; Broadway and 6th run within a block of each other north of 14 and Dekalb is...quite the dumpster fire.

If you said to me that I could fix just one thing in the system, this would be it. That junction has been studied over and over and over again, and the recommendation is always the same: fix it. This is the sort of thing that won't get easier with CBTC, and is the sort of thing that's infinitely easier to do now than it will be later. I do not overstate the issue when I say that this is *the* capacity constraint on the A division. I'd personally do Utica/New Lots and Flatbush.

Ugh. Yeah but...this isn't true. Most of the time, a hold will be for some connection or for a gap rather than to hold to time. And at any rate, holds to time affect only those trips which pass through the hold point.