Over the past couple weeks, a collaborator and I have been working on developing a more accurate way to determine standee numbers on LIRR trains. As I alluded to in Friday's Middleseataphobia post, when the LIRR conducts its passenger counts each fall, they assume that everyone will voluntarily sit in the middle and facing seats, and that is how they manage to say the vast majority of LIRR trains have zero standees. Sunny Zheng, the league leader in FOIL requests and the proprietor of TransitDocs, a website that hosts a variety of different ridership data and analysis for various transportation agencies in the Northeast, has been working on the following set of data that seems to be more realistic as to what you might see on a train. We've been bouncing this big excel spreadsheet back and forth over the past couple weeks and the synthesis of our thoughts are below. His thoughts are shown in green and mine are in black.
Every fall, the LIRR sends out an army of temporary employees to count the number of passengers on every single train, both east and west of Jamaica. Counts are often done multiple times and then that data is averaged to weed out non-representative days (days like Fridays, bad weather days, major events, service disruptions, etc., occasions where the ridership numbers are either much higher or much lower than they are on a normal, run in the mill day), and then compiled into the annual Ridership Book. The book is then further condensed into a Ridership Report that is presented to the LIRR Operations Committee at its March meeting the following year, and either makes the bosses happy or gets them worried.
But it doesn’t end there. Those counts that make up the Ridership Book eventually determine how many cars get assigned to each train, how many “programmed” (expected) standees there will be, loading guideline adherence, which trains aren’t being used much and might be cut the next time the budget is strained, or where an extra train might be added when funding becomes available. From this, the LIRR can determine a loading guideline for each train. The loading guidelines are percentages that represent the ratio of riders to available seats on the train. Those loading factors can be as low as 0.95% for a couple of early morning trains that have the same number of people as cars on the actual train to as high as 103.48% of the LIRR's most crowded train, train 2078, the 6:21pm train from NYP to Ronkonkoma. Having balanced loading factors is very important. If the loading factor is too low, then the LIRR is needlessly towing extra cars around that waste equipment, electricity, and crew members. If the loading factor is too high, then you have overcrowding situations that can become pretty dangerous if the slightest thing goes wrong. As a general rule of thumb, loading factors of 60%-75% are best for off-peak trains and 85%-95% is considered acceptable for peak trains. Anything over 95% (and there are currently six trains that exceed 95%) is deemed unacceptable by the LIRR.
There is also an annual event that is chosen for study, to further examine the ridership patterns of a particular market. (For 2012, this was Barclays Center event ridership). Given the importance of this document, and the fact that the Report presented to the Committee is highly general in nature, I decided to obtain the ridership books for the past few years under the Freedom of Information Act.
There are a few anomalies and misrepresentations in the data, but for the most part the counts are in fact accurate given the conditions stated. One key issue is that intermediate ridership, whether east or west of Jamaica is not included. Perhaps the most visible manifestation of this is train 666 (the 5:52pm from LI City to Port Jefferson), whose most crowded point is not upon departure from Jamaica, but rather upon departure from Hicksville, where it receives hundreds upon hundreds of passengers from 1258 (the 6:01pm flyer from Penn Station). Trains that stop at Jamaica are counted immediately before their arrival and immediately after their arrival at Jamaica, so a rider traveling from Babylon to Rockville Centre on the Babylon Branch would not be counted.
Another issue is that arises is the fact that consists are not always as long as “scheduled”, which is a lot more common than official statistics indicate or those “running with fewer cars” announcements might dictate. Both of those instances only occur when the train length is so short that not enough seats are available for the number of passengers listed in the book. The former case is rare enough that we need not consider it when looking at the system as a whole, the latter is really a matter of equipment reliability and maintenance and not a scheduling determination.
Since the book doesn’t show the crowding of trains, I used the given consist lengths and the number of seats per car to calculate the load factor, or how many seats on the train are filled (if the loading factor is greater than 100% that means there are more passengers than seats). For this, I used the actual seating capacity of M7 and C3 equipment (assuming that all C3 trains with 6 or fewer cars have a cab car and that there are an equal number of trailer and toilet-trailer cars on C3 trains). There are considerably more M7’s than M3’s in the electric fleet, and it is impossible to ensure that M3 equipment can be assigned to any particular run on a regular basis, so all of the factors for electric train are determiend based on the seating capacity of the M7 equipment. An Excel table showing the scheduled ridership, number of cars, and loading factors for all of the LIRR's trains. You can switch between days and directions using the options at the bottom of the table below. True to the LIRR’s word, only a very small handful of rush hour trains have standees (some small standee values show up as 0 in the official book because I used more precise seat counts). Some trains come close to that 100% tipping point, but most, even during the rush hour, don’t top 90%.
To view the entire spreadsheet in a separate window, click here.
To most commuters though, this sounds like an absolute lie, since one can often see many standees on many trains. But unless you happen to ride one of the 6 most crowded rush hour trains, the LIRR considers your train standee-free on the average day. The secret lies in the middle seat. Last year, I decided to count how many of these “undesirable” seats (middle seats, facing seats, fold-down seats) were in an M7 pair and each type of C3 car. The net number of “desirable” seats is shown in the “end seat” column. As you can see, the difference is significant, with a 12-car M7 set losing 32% of its capacity without the seats that nobody wants. After recalculating the loading factors to use the number of end seats instead of total seats, you can see loading factors that are more representative of what one might actually see. Tons of trains in the rush hours have 100% end seat loads and even a handful of off-peak/weekend trains brush 100%, all the while the number of supposed middle seats taken is in the hundreds.
Since quantifying the number of standees on an entire train based on looking around one’s own car is somewhat difficult, I decided to break down the data into a per-car basis. This is the data indicated in the “Pr. w/o e. seats/car” column. I hypothesized that since a few middle seats/fold-down seats were taken by groups or intrepid commuters, but a few end seats would be taken up by sleeping passengers or their belongings, the resulting calculation would be a fairly accurate representation of the real number of standees per car. Based on the few rush hour trains I frequented, they seemed right. But one’s own experiences aren’t always a sound judgment. Plus, the breakdown assumes even distribution of passengers along the length of the train, which isn’t always true.
Two Fridays ago (the text is black now), during the same stay that spawned the Middleseataphobia and Track Calling Patterns at Penn Station posts, I spent some time at New York Penn Station during the rush hours observing seating patterns and taking rough standee counts. From what I saw during the first half of the rush hour, the passengers without end seats per car numbers in the table above seem to be a very good accurate of what you might expect to see on any given train. The numbers do a pretty good job of balancing themselves out too, as additional occasional riders that are just riding for the day are often balanced out by those who sit in the middle/facing seats. The numbers still aren't perfect, but it's miles better than the 0's the LIRR keeps putting out.
Just for fun, we decided to quantify exactly how many cars would be needed on each particular train to give every passenger an end seat. During the rush hour, the results can be quite amusing, with many rush hour trains requiring 14 car electric consists or longer, with one particularly bad Ronkonkoma train asking for 20 cars! Other trains during the off-peak and weekend periods could probably get along with fewer than the current amount of cars. Electric cars come in married pairs (odd numbers are rounded up) and can be run in sets of no less than six cars and no more than twelve. In some instances, trains are already as long or as short as they can be, but in some instances, they could be longer. While the last column indicates how the consists should be adjusted to give everyone an end seat, that does not take into account things like yard space, equipment rotations, etc., so while a train might have "longer" or "shorter" indicated in the columns, it's probably not that easy to simply tack a couple cars on or take a couple off, so take those indications with a grain of salt. In a perfect world, the LIRR would be able to actively add and drop cars in yards and on platforms during the off-peak hours and run only 6 or 8 car trains where the loads don't need a full 12 car set, but that is easier said than done. During rush hours, pretty much every MU train is either long enough or needs to be made longer to accommodate everybody. There aren't too many instances where they can actually shorten rush hour trains.
Then there is the east of Jamaica/west of Jamaica disparity. Changing at Jamaica is a fact of life on the LIRR, since it is impossible to provide equal service to all three terminals at all times. There are also large ridership disparities among the different terminals. New York sees a lot more passengers than Brooklyn and Brooklyn sees a lot more passengers than Hunterspoint Avenue. There is no perfect way to balance service to all three of these terminals since there are people who are trying to get to or from all of these terminals at all different times of the rush. Therefore, lots of trains have to bear extra connecting burdens west of Jamaica. Trains from places like Far Rockaway or Hempstead can often arrive or depart NYP absolutely crush loaded but then half of the load empties out at Jamaica to connect to other trains.
Connections makes it tough on the LIRR since a train might require a full 12 car set west of Jamaica to accommodate all of those connecting passengers, but now once it gets east of Jamaica the load only really needs 6 or so cars wroth, and you have these extra six cars and two or three crew members that have to go all the way to the eastern terminal and take up more space there. It's a difficult conundrum and there's no real solution to it. Beefing up service along the smaller branches so they are running as many trains as possible would help, since you would be able to distribute the connections more easily (i.e. you wouldn't have one train having to support 3 trains worth of connecting passengers you would have three trains supporting one connection each.) When East Side Access will make things slightly better, but if an imbalance between the two terminals starts to appear (and I think there will definitely be one), and the service levels do not match that imbalance (i.e. trains to NYP and GCT are split 50/50 while the ridership is only really split 65/35) there could be even more trouble. Also, the LIRR has been flirting the idea of getting rid of timed connections at Jamaica, which can be very dangerous at first, for reasons I explained here.
In conclusion, the LIRR is very busy. Very busy. The standee counts they report are not incorrect, but are very misleading. As the LIRR stands right now there is little they can do to actually improve this. Penn Station is at capacity, the Mainline is at capacity, eastern terminals are at capacity, and the Babylon Branch is at capacity for lots of the rush. The LIRR can add more trains on the smaller branches like Long Beach, Hempstead, West Hempstead, and Far Rockaway and use those to push connecting loads towards Brooklyn and Hunterspoint Avenue, but there is only so much they can do. The completion of East Side Access will help, but only if the eastern capacity improvements are completed as well. In fact, if most of the ESA readiness projects are not completed by time ESA is finished, the standee situation can actually become much, much worse for the first couple years until the LIRR exactly figures out what terminals should get which trains from which branches. Hopefully that is something LIRR management has in the back of their mind. East Side Access is coming, and if the LIRR is not ready for it east of HAROLD interlocking, there's going to be big trouble!
No comments:
Post a Comment