New York’s Subway Frequency Guidelines Are the Wrong Approach

I don&#x27;t think the MTA is aiming for optimal scheduling. Peak demand depends on two things: the number of trainsets available, and the track capacity. I recently read their in-depth analysis of the A&#x2F;C lines. A&#x2F;C capacity is strictly limited by the interlocking between Hoyt-Schermerhorn and Jay St - Metrotech. It can handle 28 trains per hour if everything runs perfectly, and they have approximately that many scheduled. Similarly, people complain about the C being 8 cars instead of 10 (actually 480&#x27; vs. 600&#x27;). Unfortunately, the MTA doesn&#x27;t own enough railcars to be able to run the C with 10 cars.<p>Off-peak service is a disaster and the only thing that explains their scheduling is that they can&#x27;t afford to pay the conductors and train operators. Their guidelines are definitely shaped by &quot;how can we keep the system running with the atrociously-limited amount of money we get from the state&quot;, rather than &quot;there&#x27;s no possible way to run 28 trains per hour 24&#x2F;7&quot;.<p>Waiting 20 minutes for a train is still better than systems that shut down at 11pm and reopen at 6am, though.<p>Edit: here&#x27;s the analysis that explains all of this better than I can: <a href="http:&#x2F;&#x2F;web.mta.info&#x2F;nyct&#x2F;service&#x2F;pdf&#x2F;AC_LineReview.pdf" rel="nofollow">http:&#x2F;&#x2F;web.mta.info&#x2F;nyct&#x2F;service&#x2F;pdf&#x2F;AC_LineReview.pdf</a><p>Great document.

Synchronised frequencies sounds bad. You want non-synchronised to avoid amplification and ensure error correction, surely?<p>If you have just one temporary bottle neck screwing up the relative distance, causing a capacity shortage down the line you will have that pattern repeating if things are synced, whereas if they aren&#x27;t, you will error correct naturally.

Terrific analysis. What is interesting to me is all of the items could be explained by a single factor: distributed planning architecture, such as a planning system that calculates each line frequency independently. Essentially each train is its own functional agent.<p>This could cause the kind of data in the article, while also having a legitimate reason for use. Independent units can be better for fault tolerance, emergency recovery, and streamlined risk management, if anything goes awry.