Montreal is converting a commuter rail line into a fully automated metro.

33 Replies to “Weekend open thread: Montreal’s new REM train cars”

  1. Speaking of new trains, yesterday afternoon I caught my first glimpse of a new Siemens S70 on a burn-in run in Rainier Valley. Nice big windows on those new trains!

    1. The REM project seems pretty innovative in a few ways:

      1. The project combined an existing EMU commuter line needing modernization with use of a new Champlain bridge that was built intending to add rail and a badly needed tie-in to the airport which is near the existing trunk line. Important pieces combined to make a nice project. By the routing existing tracks or anticipated tracks, it’s ability to get quickly past debate and shovel-ready happened quickly.

      2. The innovative guaranteeing of the loan by their public employees pension fund (CDPQ) prevented a messy public funding and repeated “value engineering” discussion. I am not sure if pension funds here can be mobilized that way — but man it could really be a huge solution for ST if they could. This really helped get the project fast-tracked financially.

      3. Their vehicle decision especially having driverless trains is inspiring and shows why looking at the latest and greatest technology applications is important. Notably, Montreal is departing from having only a rubber tired Metro line for light rail.

      4. The execution of this project has going so well and so quickly that a number of additional Montreal rail transit expansion projects are now in discussion or development after a few decades of expansion stagnation.

      The project would be like what would happen if the Downtown train tunnel could primarily be for passengers and the South Sounder line was electrified and the design was separated enough from crossings to have safe driverless operations and a new ship canal bridge was built with the intent of carrying rail. I realize it’s pretty far-fetched and that lots of prior choices enabled this, but applying this concept as a Puyallup to Interbay or Ballard project using the existing rail tunnel in an existing corridor owned by the public and having this open in less than 10 years would be amazing.

      More profoundly, it shows how our institutional and incremental boxes work against expanding our rail transit network in quick and cost effective ways.

      What ideas should we take from this project? Anything?

      1. Try not to take too many ideas from the REM model. You can read of its many problems from blogger Anton Debrau at, but I’ll summarize here.

        While the project planners have made decisions that help boost ridership on this line, they work for a private entity, and are working against the overall public interest.
        1. The EMU line the REM replaces, while it had low frequency imposed by it’s commuter rail fares, had high quality track and 25kV electrification which they have completely demolished down to the bottom of the track bed. The REM solution cut off another existing rail line and precludes any further expansion of transit or inter-city trains through the Mont-Royal tunnel.
        2. The selling of the CDPQ as a ‘semi-public’ entity gave it the political capital to ignore the negative assessment of our regional planning authority and to severely undervalue existing public assets that they purchased.
        3. Several decisions for operating the REM run counter to transit goals, instead benefiting other CDPQ investments. Notably, the west island branch runs to several low-density and greenfield locations (A-40 corridor) instead of serving the existing population centers further north (Hydro corridor). CDPQ is building a half-billion dollar extension to the airport, tunneling under the biggest wetland on the island when another existing rail line runs within 1km of the airport terminal (but that also terminates 500m further west in downtown, further from most hotels). The future system capacity is also strongly limited by the choice of technology.
        4. We still have no idea how fares for the REM will be sold to the public.
        5. It’s not clear that the level of capital investment demanded by CDPQ is sustainable. With CoVID, the airport had to pull out their $400 million contribution to the airport link. There’s discussions ongoing for the city to bail out the link, despite a design that would fail an alternatives analysis.

      2. I have to say that rail projects in other cities often look more attractive than those at home. I think it’s healthy to understand that this project is not perfect.

        Also, other rail projects have lessons on what to do as well as what not to do. Because REM is a departure from “business as usual”, it is more intriguing to consider as a case study.

        Thanks for setting up some of the issues! I hope others can take a look and make both good and bad observations.

        I think the key baseline is that the existing commuter line was already electrified — and that Montreal has a rubber tired metro so this is a technology switch — and that Montreal’s harsh weather creates a different perception of rail transit.

      3. The challenges in making surface MLK driverless would seem to be twofold:

        1. Respond properly to the four or five cross streets. The existing stop/go signals are designed to do that, and it seems like it would be easy to put sensors at those intersections to make trains respond to the signal, or to stop if something is unexpectedly on the cross street.

        3. Respond to people illegally walking or driving onto the tracks along the rest of the alignment. This would require more investment than #1 because it’s so many more miles. But it would certainly cost much less than wiring all the streets in Rainier Valley for driverless cars. Especially since Link has a control center monitoring the trains the whole time, while driverless cars don’t.

        Speaking of which, does Link have positive train control or equivalent so trains can be stopped remotely if the driver doesn’t? I assume the control center can do that.

      4. From a technical standpoint, Justin has it right. The train operates remotely in Rainier Valley. From Mount Baker to Rainier Beach takes 9 minutes. It becomes grade separated soon after that, which means that the area that needs supervision is about 10 minutes or so. Since you can only have one train every six minutes, theoretically you could operate it with two remote drivers each direction (during rush hour). You would probably want to push that up to three, maybe even four (depending on how long you operate with six minute trains). That gives drivers a chance to take a little break and get ready for the next train. So maybe 8 drivers total during rush hour — or better yet, all day long.

        In comparison, Lynnwood to Federal Way will take 73 minutes, or 12 buses one direction (peak). That means 24 drivers not counting break time. With break time I figure it is about 40 drivers. If the train is running every 12 minutes you can get by with half that (20 drivers).

        This is a huge savings. During peak, you go from 40 to 8. Outside of peak, you go from 20 to 8, even if you decide to run the train as if it was peak. There are plenty of other costs, but this would still be a significant savings.

      5. The technology already exists ( The only reason for a a driver to remote in for 10 minutes is consumer acceptance or unions.

        Will there be several centrally located drivers ready to take over if the train encounters a unique situation? Sure. But 99% of the time the trains should be able to cruise through RV with zero human supervision. If something – person, vehicle, a large turkey, whatever – wanders on to the tracks, an automated train will react much faster than a human driver, remote or in person.

        We will need bus drivers for a long time, but we can convert all of Link to driverless whenever ST decides to make the capital investment. Buses don’t follow a perfectly prescribed route but need to deal with random deviations almost every run; trains do not. Bus drivers do far more than just drive – they enforce fares, ensure passenger safety, etc. – while Link drivers simply drive and open doors.

  2. Is there any possibility of automating Link lines in the future? I could see at-grade crossings along MLK posing a problem, but perhaps it’d be possible for a remote operator to take over if there was an issue.

    1. The biggest problem is political, ST hasn’t considered it important. And it’s assumed the unions would be against driverless trains. In the current alignment driverless trains would require eliminating the level crossings on MLK or some emerging technology. ST talked about exploring how some lines in ST3 might be driverless but it never pursued it. Just like how its earlier long-range plan had an item to consider 24-hour lines someday, but it never pursued, it and now it’s gone from the plans as far as I can tell.

      1. Whatever LIDAR technology is used to keep self driving cars out of the way of pedestrians could be equipped today on Link trains for MLK. The fact that the trains travel a known, fixed route makes the technology much simpler. If the software encounters a shadow it doesn’t recognize, it can simply stop; a human in the control room can then manually inspect the cameras and, if it’s nothing, cautiously move the train ahead, remotely, until the computer is able to take over again.

        The real obstacles to doing this are two-fold:
        1) Public agencies are very risk-averse. Nobody wants to be the first to do it, lest they get sued if anything goes wrong.
        2) The drivers union will not stand for their jobs being automated away. If ST tried to automate Link trains, they might find themselves without bus drivers or mechanics. The politicians that make up the ST board also don’t want to incur the wrath of the drivers’ union in their re-election campaign.

        Technologically speaking, automating Link trains is easier than automating Uber cars. Politically speaking, automating Uber cars is much easier because Uber is a private corporation, not subject to unions or politics.

        Because of the politics, I can easily see the Link trains continuing to have human drivers decades after the taxicabs are mostly automated (and, if you want a human driver, you have to pay extra for a premium service), similar to how the New York subway still runs their trains with both a driver and a conductor, decades after that was no longer necessary.

      2. @asdf2,

        I’m not sure your assessment is correct. Ya, politics matter, but unions aren’t as strong as they used to be, and even our deep blue local reps aren’t as tied to labor as they used to be. And certainly our local electeds still feel pressure on the fiscal responsibility side of the ledger.

        You are correct though that automating Link is a much simpler task than automating a taxi. Being on a fixed guideway eliminates lots of variables, but it is more than just that. Each Link vehicle is also tied into a pretty extensive SCADA system which also would potentially simplify automation.

        And it is not just the economics of replacing the driver that favors Link automation over taxi automation. This tech will be expensive. As such, it is much more likely that someone like ST would spend the $50k (just a notional number) to automate an LRV that operates 18 hours per day and carries 200 fare paying people as opposed to the taxi owner spending the same amount on a taxi that only carries 3 people.

      3. Aren’t the two big issues with the Rainier Valley segment that a) SDOT has wanted to preserve the current allocation of time set by traffic signals for cars crossing MLK and b) cars keep colliding with the Link trains?

        If cars are automated, it should help with the Link reliability problems caused by (b), and you could maybe come up with a story where automating cars leads to SDOT allowing for more frequent signal changes along MLK, which would allow Link to run more frequently, but I don’t see how automating Link would allow for much greater frequency in the RV segment.

      4. @PhillipG — It wouldn’t be better peak frequency. It could lead to better frequency outside of peak, which is the vast majority of the day, and when the vast majority of people use the train. In other words, they couldn’t run the trains more often than every six minutes, but it would be a lot cheaper to run the trains every six minutes, all-day long. For that matter, it would be cheaper to run the trains during rush hour too.

        There are other expenses though, but this could lead to more frequency in the middle of the day.

    2. It will probably go along with self-driving car technology someday. Driving a train would be simpler since a train only needs to speed up and slow down, with no steering required. The Montreal Metro is at that stage of automation, where there is a driver but they only control the doors and tell the train when to leave the station.

    3. I agree with Lazarus, I don’t think the biggest problem is the unions. I think there are two big problems:

      1) Financial. This would have a long term payoff, but would require spending extra money right now. ST’s financial situation encourages the opposite.

      2) Motivation. The big advantage of an automated system is that it makes it more cost effective to run the trains more often. You still have maintenance costs, but you don’t have driver costs. The problem is, ST doesn’t seem to care about frequency. When they comment about frequency, it is only in the context of crowding, not ridership. For example, there is no talk of just running the trains every six minutes, all day long, once Link gets to Northgate. It would be trivial to calculate the extra ridership and extra fare revenue. Presumably ST knows the costs. It would therefore be quite easy to release a report that says “running trains all-day long every six minutes would gain this many riders, but cost this much”. But they don’t. They don’t even seriously consider it. They just don’t care.

      They aren’t focused on ridership — even cost effective ridership — they are focused on building new things.

      1. To say that Sound Transit just doesn’t care about frequency is an exaggeration. If they didn’t care, Link wouldn’t be running every 10 minutes all day. It would be more like every 30. A train every 10 minutes all day is comparable to Washington D.C. and better than many transit systems, including the San Francisco BART and the Portland MAX. So, obviously, they do care about frequency somewhat.

    4. For the last 60 years, you people couldn’t figure out how to automate the one mile long monorail.

  3. When are we going to see trains at Northgate? I thought ST had to run non-revenue service for 6 months prior to opening.

    1. Link is closed north of SODO this weekend and the next few weekends, so they’re doing something.

      1. They are making the final OCS and signaling/com connections between the old and new systems.

        They will undoubtably have to do the same thing again in 18 months or so to connect up E Link.

      2. I believe they are taking the first steps to install overhead electric for East Link at IDS.

      3. @Jimmy.

        Ya, I believe you are correct. This is for an E Link connection. That is even better news!

    2. I saw one of the new Siemens trainsets parked just north of the Northgate station last night so I’m guessing testing has now begun.

      1. I also saw that this morning. I drove past Northgate at 7:10am this morning and saw two 4 car Siemens trains side by side at the pocket tracks just north of the station with interior lights on. This is the first time I have seen the new trains in that area. Two weeks ago I saw train activity at Northgate, but it was the original Link trains.
        On a different topic, I usually look the other way to see the advancement of the pedestrian bridge. Since I saw the trains, I missed what was going on the other way.

    3. I replied to a comment by Squints but I think it will answer part of your question. See below.

  4. Fire destroys building under construction on 23rd and Plum Saturday morning. I just binged that address. Is that where a 133 unit congregate housing complex was being built?

  5. Montreal’s REM is an example of a how Rapid Transit proposal just doesn’t die, it Mutates. Back in the 1960’s, it was supposed to be Line 3, and there are other proposals that pre and post date that period.

  6. Honestly doing an RER system for Montreal like they are for Toronto with GO Transit. would’ve been a far better idea. Sure it may take significantly longer to do but it’s worth it. Cutting off Montreal’s only mainline is a big no-no and a death sentence for VIA HFR or HSR up in Canada. They’re gonna have to build a second tunnel under Mount Royal and considering North American construction costs, that will surely be expensive. And like other have mentioned, they work for a private entity, and are working against the overall public interest.

  7. No surprise, the Seattle Times editorial board likes fast passenger ferries. “For commuters, that’s an extra hour to sleep in… That could open up a lot of affordable housing without adding traffic to already-congested Seattle-area roadways. The median home price in Kitsap County is just over $461,000, compared with around $723,000 in King County, according to the Zillow Group. Everybody wins.” Luckily foot ferries are high capacity and require less subsidies than surface transit. Oh wait.

    And calling $461,000 affordable is an insult to half the population.

    1. Also, there’s a very limited number of homes available near the ferry. If more than a tiny number of people move there, they’ll be bidding prices up in Kitsap county and driving increasingly long distances just to get to the ferry dock. On top of that, even with all those subsidies, the fares for foot ferries are expensive – I believe around $15-20 round trip/day. By the time the cost of commuting is added in, your not that much better off financially than just find a place to live in Seattle and having a short commute.

      1. Bremerton wants to grow, significantly. Build more housing around the ferry docks. Bidding up prices is kinda the point of the foot ferries, from an economic development standpoint.

        Also, the levy for the foot ferries also funds feeder bus service. The foot ferries provide a strong anchor to support good bus service (albeit commuter oriented), which seems useful.

Comments are closed.