Followup on the Providence Regional Rail Shuttle
Peter Brassard’s proposal for a very frequent-stop mainline train in Rhode Island received comments both here and on Greater City, dealing with issues from rolling stock to station choice to scheduling. Some are fairly trivial, some aren’t. The upshot is that the project is technically feasible, but requires political head-bashing, especially with regards to scheduling.
First, the easy part: if the line is only to run between Central Falls and Warwick, then the rolling stock should be electric; this both improves performance and eliminates a political bottleneck, because the EMU market is larger than the DMU market, and in case FRA regulations do not change and obtaining a waiver is too expensive, there are M8s ready to use. The M8s are heavier than is ideal, but their performance is to my knowledge imperceptibly worse than that of noncompliant trains in the speed range appropriate for the short stop spacing, up to about 100 km/h.
Scheduling is the problem, because there has to be track sharing with something. The line is three-tracked: there are two tracks for Amtrak, also used by the MBTA north of Providence Station, and one track for freight. The line used to be four-tracked, but was reduced to three tracks in the 1990s in order to widen the track centers and allow the Acelas to tilt. Further reduction in track centers is not acceptable: at 4 meters (more precisely 13′) the distance is shorter than the standards for greenfield construction in Europe and even Japan. Track center standards are laxer on lower-speed segments, as the trackage through Providence is, but tilting becomes unsafe for an Acela-wide train. (The Pendolino is 37 cm narrower than the Acela.)
The alternative is to slightly widen the right-of-way at certain overpasses to allow four tracks, for a minimum of 20 meters with 4-meter track centers; some work, including widening, is already required to make room for platforms, and many of the most constrained locations, such as Olneyville at 18 meters, are station stop sites. It’s this construction that would most likely be the bulk of the project cost. At much lower cost, it would also allow electrification of the full corridor, making EMUs a feasible rolling stock choice for the local trains.
With four tracks, the question becomes, what regional rail should share tracks with. The choice is between intercity trains, which are currently slow but could be sped up, and freight trains. Both require political maneuvering, because neither Amtrak nor the Providence and Worcester has operating practices that are compatible with punctual passenger service. (Amtrak is more easily reformable, but an Amtrak that’s been so reformed is an Amtrak that runs trains much faster on the Northeast Corridor, increasing the regional/intercity speed difference).
I contend that it’s actually more correct to share tracks with freight. The sharpest curves are at stations, and so no superelevation is needed, but even if it were, allowing 100 km/h passenger trains could be accommodated with minimal freight train cant excess (about 25 mm at 50 km/h). More importantly, freight and local passenger rail have similar average speeds. The speed profile is different – freight is steady and slow, local passenger rail attains higher speeds but makes frequent stops – but when headways are long enough, this is not a problem.
On page 46 of the Providence Foundation study on a similar passenger line, we see that there aren’t many freight trains, so headways are determined by passenger trains. The freight schedule on page 48 of the same study suggests that freight and passenger train speeds would be very similar. It has trains doing Pawtucket-Warwick in 23 minutes; modern EMUs with a top speed of 100 km/h (losing 45 seconds to each station stop) and making the proposed stops would do the same in 25 minutes, with 7% padding. The local passenger train is a hair faster than the freight train on the Providence-Pawtucket and Cranston-Warwick segments, in both cases by less than a minute, and a bit slower on the Providence-Cranston segment, where station spacing is denser. This is close enough that I believe that 15-minute passenger train frequency is no barrier to track sharing. Potentially even 10-minute frequency can be accommodated. It requires freight trains to be somewhat timetabled, but they’d have a window of several minutes to enter between each pair of successive passenger trains, and missing their window would not delay them by more than 15 minutes. There is, then, no technical barrier to sharing tracks with freight.
The alternative, sharing tracks with intercity trains, is more dubious. Although less construction is required, the speed difference is larger. Instead of taking 23 minutes between Pawtucket and the airport, optimized intercity trains would take 8:45, including padding and a station stop at Providence. They can pass local trains at Providence, at the cost of slowing them down by several minutes while they wait to be overtaken, but even between Providence and the airport, travel time would be 5 minutes for intercity trains and 17 for regional trains.
If there’s four-tracking in Warwick, or two stops are dropped, then it’s tight but doable. Otherwise, it’s not; 12 minutes is too long a window for 15-minute service. It would require an extra terminating track at Warwick, but that would be needed anyway. The problem then is that local Rhode Island trains and MBTA trains would interfere with each other at Providence because both would dwell at the station for too long.
Interlining the two services and having MBTA trains make local stops in Providence is possible, and in conjunction with the two-overtake schedule for Boston-Providence naturally yields a three-overtake schedule. The problem is that the more overtakes there are the more reliability suffers. If an hourly freight train misses a window and needs to be delayed 15 minutes, it’s no big deal; if the goods couldn’t take a 15-minute delay, the train would be sufficiently punctual to make the window. If a passenger train misses a window, it requires the train behind it to slow down and this is not recoverable if the schedule is so tight.
When it’s unavoidable it’s best to just invest in running trains on schedule, but in this case a three-overtake schedule is avoidable. Thus track-sharing with freight is the correct option, leaving intercity trains to have a track that’s entirely theirs south of Providence, as this shuttle concept would almost certainly take over Wickford Junction service if necessary. It conveniently also allows higher regional rail frequency should the need ever arise, and because the scheduling is loose makes it easier to shoehorn another line into this system.
The parts of the proposal that involve sharing tracks with freight are superficially similar to the North London Line, which successfully mixes 4 trains per hour of suburban rail service (using four-car EMUs) with freight. I believe that during rush hour there are 6 trains per hour and no freight. I agree that mixing with long distance trains is less desirable, at least if you want 15-minute headways. There’s also the issue of reliability: sharing tracks only with freight makes the system fairly isolated from delays elsewhere, while sharing with MBTA exposes you to delays in Boston, and Amtrak exposes you to delays on Metro North, and vice versa. With 30-minute headways, it might be feasible to share with the existing intercity service.
The M8s are heavier than is ideal, but their performance is to my knowledge imperceptibly worse than that of noncompliant trains in the speed range appropriate for the short stop spacing, up to about 100 km/h.
Any of the post WWII EMUs on the NEC could and can do 100MPH/160KPH. They don’t when they are running local but they can.
The claim is that below 100km/hr (certainly a relevant speed range for this stop spacing), the PERFORMANCE of that compliant EMU is not appreciably worse than the performance of non-compliant EMUs (whereas the performance of compliant DMUs IS appreciably worse than the performance of non-compliant DMUs).
There is nothing in the statement that could be read as saying anything about the top speed of the EMUs, so remarking on the top speed of “any post WWII EMU running on the NEC” would be a non sequitur.
Being able to go 100 comes in handy when the station spacing – when they decide to extend every third train to New London lets say – isn’t under two miles/three kilometers.
Does the freight on the NEC run much faster, and with shorter trains, than freight on the west coast? There is a Union Pacific line right next to my light rail line, and only trains of empty containers go fast. The fully loaded trains seem to go about 20 to 30 mph, and they are about 1 mile long. Could they really fit in between passenger trains with 15 minute headways?
Likely yes, as the dominant traffic is passenger. As far as Rhode Island is concerned, there probably won’t be much of a problem dealing with any heavy freight activity, as Providence and Worcester seems to have most freight activity on the Shoreline further south in Ct. between Old Saybrook and New Haven.
Not sure about freight in general, but the P&W trains on the route average about 35 mph between Pawtucket and Warwick. It’s actually a hair more than the average speed of the passenger trains proposed, because of all the stops.
How much faster does your light rail line run than those UP freights?
In the Providence Foundation study that Alon links to, they describe the P&W sets as two and three locomotives and thirty freight cars, so not mile long coal trains or transcon double stacked container train.
I’ve paced freight trains of all types on the UP mainline next to I-80 in Western Nebraska at freeway speed, so freight trains can operate fast if there is an advantage for it. Generally freight railroads don’t like to operate trains fast, then slow, then fast, etc, because it takes a lot of energy to speed up and slow down.
Heck, even 60-year-old steam trains can reach freeway speed: http://www.youtube.com/watch?v=5sXQFE2jAsw