First, let's talk about slippery rail. It is a true problem for railroads around the world - it isn't just an excuse they're creating to justify delays. The advantage of moving trains by rail - low friction & higher efficiency - becomes a problem when conditions are not ideal. The leaves that fall during autumn get crushed on the rail head and create a slippery substance. Locomotives have trouble stopping and accelerating on these slippery rails. Natick Center is a notoriously bad spot, especially on track 2. Track 2 is depressed to maintain adequate clearance for tall freight cars under the roadway bridges in downtown Natick. This depression is just enough to cause problems for trains stopping at or trying to depart Natick Center. But the slippery rail can cause problems at any station stop (but note that slippery rail is less of a factor for express trains - it is primarily a problem for trains stopping and starting frequently, such as local trains).
The primary method that Keolis uses to mitigate the effects of the slippery substance from crushed leaves is the operation of two 'wash trains' - one train on the north side and one on the south side. Each train is essentially just a power washer that cleans the substance off the rails with a blast of 15,000 psi water. The power washer is mounted on a flatbed railcar and a tank car is included in the set to provide a water source.
Old announcement with photo about wash train here.
Article about world-wide troubles with slippery rail here.
Keolis video about slippery rail here.
Keolis flyer about slippery rail here.
I have inquired about the usage of the wash train on the Framingham-Worcester line, but I haven't received any confirmation of when it last operated on our line. Certainly the strong wind storm over the weekend which blew down many leaves didn't help, and it may have been too late to deploy the wash train to mitigate the damage done by the storm.
As you would guess, slippery rail prevents proper traction between the train wheels and the rail. In addition to preventing the train from stopping or starting normally, damage can occur to both the rail and the locomotive.
Wheel slip can actually occur in two different scenarios: stopping and starting. When a train attempts to stop and the wheels slip, some wheels STOP rotating and the train will seem to lurch as you enter a station. You may be familiar with this sensation - it is apparent in extreme cases. It is a very similar situation to a car's brakes 'locking up' when skidding on ice or snow. And also similar to a car, when traction motor wheels slip during acceleration and departure from a station, some wheels may SPIN after losing traction against the railhead. Here is a crazy video of extreme wheel slip on a freight train (although this is probably not related to slippery rail, it still illustrates what happens with wheel slip).
Below is a photo of damage to a rail at Grafton station on our own Framingham-Worcester line caused by wheel slip - a traction wheel on a locomotive spinning but not moving the train anywhere. Kudos to one of our alert Twitter users for spotting this! Note that this photo is from a year ago - not this season.
@ehthayer wow. that's pretty serious. wheel slip damage. @MBTA_CR
— Dave (@FramWorMBTA) November 17, 2015
P500 ended up arriving in Boston about 60 minutes late. In the meantime, that jammed up the trains behind it. P502 was converted to a local (rather than an express) - although I'm not sure why they did this. But the conversion to a local had no major effect on the on-time performance of P502, since it was already blocked by the very late P500 directly ahead.
P580 & P504 were delayed by the traffic jam of P500 & P502. P506 and P508 both experienced minor delays, but only as a result of their own problems dealing with wheel slip and slippery rail - not because of the traffic jam from P500.
So by 7:00 - 7:30 AM, most of the residual delays from the very delayed P500 were being worked out. While inbound trains were still going to be delayed by the continued slippery rail conditions, those delays would have been relatively minimal (P500 effectively cleaned the track as the first train across the line).
This is where the other factors of the morning start to play a role in the whole story.
The second factor is the "equipment cycle" or "equipment rotation." The Worcester storage yard only has space for four train sets. The four trains that are stored there overnight are used for P500, P502, P504, and P508. Every other train on the schedule is a 'turn' of an outbound train. Long time readers are familiar with my frequent references to the "Turn Table" which explains this concept in detail.
With the new schedule implemented in May 2016, most Framingham-Worcester trains remain on the Framingham-Worcester line all day long - going back and forth between Worcester, Framingham, and South Station. So this is the second factor - with trains P500, P502, P580, and P504 all delayed on their inbound trips, their subsequent outbound trips - and then potentially the next subsequent inbound trips - were bound to be delayed. I refer to this as 'cascaded' delays.
The normal equipment rotation at South Station includes the following 'turns:'
P500 goes outbound as a 'deadhead' / non-revenue super express direct to Worcester, where it becomes the P552 "Heart to Hub" bullet train.
P502 goes outbound as P585 which then turns at Framingham to become P584.
P580 goes outbound as P505 which then turns at Worcester to become P512.
P504 goes outbound as P587 which then turns at Framingham to become P586.
Therefore, there was already a potential for P584, P552, P586, and P512 to be delayed.
Here is where the third factor comes into play. At about 7:45 AM, the Keolis dispatcher instructed train P585 to continue west as P505 and go all the way to Worcester and become P512. This actually made some sense to me - P585 operates ahead of P505, the usual equipment for P512. Having the P585 equipment get out to Worcester could mean less delays for P584 and P512 - assuming the P505 equipment would go out to Framingham and operate as P584.
Keolis sent out a message at 7:32 AM that P584 would operate 10-20 minutes late. This was NOT surprising to me, given the 'cascading delay' issues noted above. P584 is scheduled to originate at Framingham with a 7:50 AM departure - just after the passage of P508. It follows P508 making all the local stops. P584 typically has ridership of ~750 passengers.
However, equipment for P584 never appeared in Framingham. Multiple requests for information from @MBTA_CR on the status of P584 went unanswered. A message went out at 9:00 AM that P584 was cancelled - 1 hour and 10 minutes AFTER the scheduled departure of the train from Framingham. There were no official messages regarding P584 between 7:32 AM and 9:00 AM. This fourth factor in our list - failed communications - is baffling, since as much as they get dumped on by lots of riders and twitter users, they are usually pretty good at major messages such as those. I can only guess that the multiple changing circumstances forced important information to get lost. I hope they conduct a post-mortem / lessons learned inspection of this communication failure and take a look at any systems or processes that may be able to be improved.
If P584 had operated 10-20 minutes late, it still probably would have operated AHEAD of P510 - and probably would have swept up passengers waiting for P584 AND P510. Any resultant overcrowding on P584 could have been accommodated by P510 directly behind.
The fifth factor sealed many passengers' fate. P510 was operating with a consist of only 5 flats - a capacity of 570 seats. The normal load on P510 is 687 passengers, so it was already destined to become overcrowded. With P584 cancelled, P510 was now being asked to accommodate the load of both P584 and P510 - approximately 1,437 passengers.
As many passengers reported, P510 was filled to capacity at Natick Center and passengers waiting for P584 or P510 at Wellesley Square, Wellesley Hills, Wellesley Farms, Auburndale, West Newton, and Newtonville were unable to board P510 due to the overcrowding on it. Video from Wellesley Farms:
P510 5 flats took in 2 pax at farms because 2 pax got off. pic.twitter.com/nNfyTPH9Jv
— Issac Gardiner (@issac_gardiner) October 24, 2016
Fairmount train 750 (a prominent subject of Sunday's Boston Globe article) was briefly cancelled at 6:45 AM and then "un-cancelled" at 6:49 AM, but as I study the timing of these events, it doesn't look like that could have affected the Framingham-Worcester equipment rotation (especially the later turn of the P585 / P584 equipment).
If you haven't read Sunday's front page Boston Globe story about the Fairmount line, it is a very important story. A follow-up article has just been posted that confirms what many of us have long suspected - there is a shortage of 14 coaches versus what is needed to operate the entire commuter rail schedule. I honestly don't believe that there is any targeted denial of service to the passengers of the Fairmount line - I think that Keolis and the MBTA have been allocating equipment to the services with the highest ridership and cancelling the lowest ridership trains. But the optics of preserving service to wealthier suburbs while cancelling trains in urban areas is problematic. The bottom line remains that for at least a few weeks (perhaps months?) the MBTA and Keolis have been trying to operate a schedule with insufficient equipment. There are no indications that can be immediately or quickly remedied. The Fairmount story and the equipment shortage aspect of this story may draw more attention until sufficient equipment can be allocated to the current schedule. There is a potential that future cancellations of trains will rotate amongst all of the lines - and affect both lightly used trains and trains with heavier loads. And perhaps that is more 'fair.' But that's just my speculation.
In summary, many factors conspired to make the Monday morning commute problematic.
1) Slippery rail. Without this problem, everything probably would have been OK. It is unknown at this time if there was anything that Keolis could have done better to mitigate the slippery rail, since it is unknown if the wash train has been deployed appropriately.
2) Cascaded delays as a result of the equipment cycle. Add a spare set into the mix and cascaded delays may have been mitigated.
3) The decision to cancel P584. Utilizing a spare set for P585 / P584 could have alleviated this problem.
4) The communications failure regarding the cancellation of P584. Had Wellesley & Newton passengers been told at 7:45 AM (when the P585 equipment was re-purposed and sent to Worcester for P512) that P584 was cancelled and/or that the next train might not be able to accommodate them, they could have sought alternate transportation.
5) The small size of P510. Already under capacity for its usual load, it was forced to try to accommodate the load of P584 & P510. Even having a relatively large set for P510 probably would not have mitigated this problem - P584 + P510 is just too many people for all but the largest sets.
Finally, please remember that the crewmembers out on the trains are NOT the people that have anything to do with any of the issues or decisions above. There were many kudos given out to the crew of P500 for keeping their passengers informed about the problems that train experienced. But there were also reports of passengers becoming irate with crewmembers regarding the overcrowding or lateness issues. Keolis crewmembers are NOT our enemies. Let's bring back some civility and direct our anger at letter writing and advocacy for a better commuter rail system.