Advocacy needs to focus on eliminating slowness

That's slowness both in terms of low speed limits (especially in the middle of higher ones) and in terms of poor acceleration. I'm currently working on a personal project that involves mapping out the San Diego-Los Angeles corridor to see the benefits of reequipping the fleet with DMUs or electrifying the line as well as identifying where improvements ought to be made; the end goal is to basically put together a list that can be used for identifying the biggest bang for the buck in terms of schedule reduction (for Amtrak, Metrolink, and Coaster) and for identifying priorities with arbitrary pots of money. With that out of the way, this is what the route looks like between San Diego and Solana Beach with the current equipment (modeled from MBTA's Fairmount study and the new bilevel car braking specifications).



As you can tell, despite several miles of 90 mile per hour clearance, the speed won't reach above 80 miles per hour. There's no point to maintaining the track above Class IV, the train simply won't make it before it has to brake for a slow zone. I've not yet done the work, but I strongly suspect that a RegioShuttle making an additional stop at Sorrento Valley (MP 18.5) would probably reach Solana Beach before a Surfliner with the current equipment despite its lower rated speed limit (75 mph). 75 miles per hour in ~93 seconds instead of 256 seconds should cut off quite a good chunk of time.

The problem of course is we get things like the Chicago-St. Louis high speed rail which raises a fair number of areas to a maximum authorized speed of 110 miles per hour, at significant expense, but leaves a large chunk of the route at slow speeds (Alton-St. Louis). Thanks to low acceleration, the gains for running above 80 miles per hour aren't terribly high and make a negligible impact upon the total running time; getting rid of slow speed zones and simply increasing the speed with which a train accelerates to speed are far more important and influential.

This particular bit does seem to map out fairly well to reality, this modeled the run at 32 minute 15 seconds; Amtrak's scheduled run time is 34 minutes including dwell time at Old Town.

Also, if anyone has a copy of Union Pacific's Los Angeles Area Timetable and could forward it, I'll be able to, eventually, do this all the way up to Goleta. And I'm aware that this may very well simply be preaching to the choir, but hey, neat visual aid for the choir to use now at the very least.

Just for reference, but what are the slowdowns at miles 8-10, 14-17, and 21-24? The one near Old Town is obviously station-related, but I'm having trouble thinking of what the others are (only having hit the Surfliners a few times so far).

Edit: To be clear, I agree with you wholeheartedly...there's a lot more to be gained by fixing "bad track" segments than by seriously raising speeds on other segments (though the latter can, and I will cynically argue sometimes should, be done to get good PR). An excellent example here is the RVM-RVR segment in Virginia...the timetable indicates that trains there take 20-30 minutes (depending on the train and the day; 25-30 is standard, but one train [194, operating weekends] is slipped in with 20 minutes) to go maybe 10 miles. Consistently getting that down into the 15-20 minute range would be a substantial improvement (even though it would be expensive). Another case would be ALX-WAS (8 miles; 15-20 minutes is standard SB...I ignore NB times because of a tendency to pad NB trains spectacularly between departing ALX and departing WAS for purposes of corridor reliability). If you could reliably cut travel times on those two segments, you might have 15 minutes you could trim out of the more padded schedules.*

There are other cases of this, too, often in major station approaches...though in some more rural areas you randomly have scads of slow track (such as on the Adirondack north of SAR). There are some infamously bad approaches to both Montreal and Vancouver, though the politics of fixing those is...problematic.

*VA's times are pretty padded, too...I've been on a Meteor that was 15 minutes early into RVR, thanks to not having any intermediate stops.

I'm afraid I think the higher priority than this is eliminating delays. Eliminating slowness is important, but the most important form of it is not sitting behind freight trains.

Anderson said:

Just for reference, but what are the slowdowns at miles 8-10, 14-17, and 21-24? The one near Old Town is obviously station-related, but I'm having trouble thinking of what the others are (only having hit the Surfliners a few times so far).

Click to expand...

The drop to 40mph at Old Town is a speed limit for the switch. CP Elvira is the 8-10 area, there's an issue with curves (though there is work to alleviate some of that while double tracking). 14-17 is the detour around Miramar Hill and 21-24 should be Del Mar.

neroden said:

I'm afraid I think the higher priority than this is eliminating delays. Eliminating slowness is important, but the most important form of it is not sitting behind freight trains.

Click to expand...

Depends on how you want to do your priorities (and sitting behind a freight train isn't a concern on the Surfliner). There's slowness in scheduling (which is what I'm looking at) and slowness in meeting the schedule (on time performance, which is what you want). The latter can affect the former (padding and holding in areas to allow other trains to go by due to congestion), but also has a fairly straightforward fix (more track or improved signals). Personally, I would put the primary focus on travel time competitiveness, which means raising the average speed as much as we can; OTP is important, don't get me wrong, but it is also pretty heavily dependent on service levels which are dependent on patronage which is largely dependent upon travel time.

I'd say what is important is reliable travel time. Historically this has been a problem for example on the NEC North/East NYP - BOS, though things have improved a lot lately, to some extent by extending schedule running times. At least on corridors I think reliability is important enough to trade away some on scheduled travel time. Perhaps not to the extent that NJT does with 15 minute padding on a 1.5 hour run, but still some.

Amtrak run time: 32 minutes 15 seconds

DMU run time: 30 minutes 40 seconds even with that extra stop in Sorrento Valley.

So, if we include the same amount of dwell time (2 minutes on the Surfliner), my initial hypothesis earlier in the opening post was wrong, but only by 25 seconds.

Paulus said:

There's slowness in scheduling (which is what I'm looking at) and slowness in meeting the schedule (on time performance, which is what you want).

Click to expand...

Studies of urban and commuter rail lines show that poor OTP can cause ridership to drop by 50%, very quickly. This is true on fast lines, this is true on slow lines. Restoring reliable schedules causes the ridership to rebound -- that is to say, to double.
Now, if you've "restored OTP" by adding so much padding that the trains are no longer fast enough for a lot of people, you'll lose a bunch of ridership, but on the whole, the ridership is usually STILL better than the ridership when the schedule was unreliable. Really.

People care a hell of a lot about knowing when they're going to get there. If the trip is long in a reliable fashion, people plan around it. But if it's *unpredictably long*, "hyperbolic discounting" comes into play. People look at the 5th percentile, or maybe the 10th percentile, or maybe only the 1st percentile, performance when they estimate how long their trip is going to take.

5th percentile means, if your train is roughly on time most of the time, but 1 out of every 20 times it is 12 hours late, then the customers assume it will be 12 hours late *every time* and plan around that. The ridership results are obviously bad.

The latter can affect the former (padding and holding in areas to allow other trains to go by due to congestion), but also has a fairly straightforward fix (more track or improved signals).

Click to expand...

Yeah, then why don't we have South of the Lake built yet? Why don't we have exclusive passenger tracks on the Empire Corridor? Heck, why hasn't the Point Defiance Bypass started construction yet? For that matter, why does E Hunter Harrisons' CP keep violating the law by dispatching freights ahead of passenger trains, even though trackwork was built specifically to expedite the passenger trains?
I agree that the solutions to poor OTP are straightforward *technically*, but *politically* they seem to be extremely difficult. That's why I think advocacy has to focus on this: getting the trackwork, and crucially, the *control*, which will allow for reliable OTP. Faster schedule time is a nice side effect, because it's much easier to do that once you have *control*. Amtrak sort of had Kalamazoo-Porter fall into its lap, and it improved the scheduled time basically as a side project.

Personally, I would put the primary focus on travel time competitiveness, which means raising the average speed as much as we can; OTP is important, don't get me wrong, but it is also pretty heavily dependent on service levels which are dependent on patronage which is largely dependent upon travel time.

Click to expand...

Due to hyperbolic discounting, I would say the important number is the average speed at the 5th or 10th percentile OTP performance level. (One could quibble over the correct percentile, but you get my point.)
The Surfliner is lucky to have OTP upwards of 80%. A few improvements in OTP and it'll be good enough that average scheduled speed is all you need to worry about. Those improvements in OTP can be achieved with a few more passing tracks, because the route is mostly government-controlled.

The Pennsylvanian has 90%+ OTP and should focus on scheduled time.

By contrast, the Adirondack is running with 30% OTP. Fixing OTP is critical.

The LSL is running with 60-70% OTP. The scheduled runtime is attractive enough; the problem is that it's running late, period, and fixing that needs to be the priority.

The Wolverines had a catastrophic drop in ridership when OTP dropped (as low as 17%) a couple of years ago. OTP is still only 47%. Get OTP up into the 90% range (without adding hours to the schedule) and you'll watch the ridership skyrocket. Change the scheduled runtime without fixing OTP, and you will get absolutely no benefit.

Most Amtrak routes are much more like the Wolverines than like the Pennsylvanian. The priority needs to be OTP.

And the priority method of getting that needs to be state-owned track, because most of the Class Is have repeatedly demonstrated bad faith towards Amtrak in dispatching and/or maintenance, and so bluntly can't be trusted with track ownership. (NS has behaved itself very well since it lost the court case with NC over the ownership of the NCRR, but the very fact that that's the stick which changed their behavior should get you thinking.)

Once your local governments own the track, you can start pressuring them to focus on fixing the slow zones and getting trains which accelerate fast. If the track is owned by CP, there's no point, CP will just eat up any money you give them and give you nothing in return.

...FWIW, when I went to the hearing in Syracuse on the NYS "high speed rail study", I told representatives of both NYSDOT and the corporate consultant that they were never going to get improvements unless they owned the track, because CSX dispatching was a problem. They both said "We know. We understand this." So there's some chance at progress.

neroden said:

Paulus said:

There's slowness in scheduling (which is what I'm looking at) and slowness in meeting the schedule (on time performance, which is what you want).

Click to expand...

Studies of urban and commuter rail lines show that poor OTP can cause ridership to drop by 50%, very quickly. This is true on fast lines, this is true on slow lines. Restoring reliable schedules causes the ridership to rebound -- that is to say, to double.

Click to expand...

Do you have a link by chance?

Now, if you've "restored OTP" by adding so much padding that the trains are no longer fast enough for a lot of people, you'll lose a bunch of ridership, but on the whole, the ridership is usually STILL better than the ridership when the schedule was unreliable. Really.

Click to expand...

Cynical question: Has extra padding ever actually brought OTP back up?

Yeah, then why don't we have South of the Lake built yet? Why don't we have exclusive passenger tracks on the Empire Corridor? Heck, why hasn't the Point Defiance Bypass started construction yet?

Click to expand...

Because you guys haven't been spending the past several decades investing hundreds of millions of dollars into the rail corridors like California? Politically, I think it is in large part dependent on commuter service; that's been a major motivation for improving LOSSAN and the Capitol Corridor and explains why the San Joaquin has seen comparatively little investment in the meantime.

And the priority method of getting that needs to be state-owned track, because most of the Class Is have repeatedly demonstrated bad faith towards Amtrak in dispatching and/or maintenance, and so bluntly can't be trusted with track ownership. (NS has behaved itself very well since it lost the court case with NC over the ownership of the NCRR, but the very fact that that's the stick which changed their behavior should get you thinking.)

Click to expand...

It's not so much bad faith I think as that they've underestimated the amount of track that they need for their own freights (seriously, BNSF is less than thrilled over how screwed up the EB route is right now) and that you'll always have some degree of interference with mismatched running speeds; I have a hunch that OTP is directly related to the number of train meets and overtakes in a route; this would naturally result in terrible OTP for the long distance routes.

Paulus said:

Do you have a link by chance?

Click to expand...

I'm really damn lazy about saving links, unfortunately. So no. These were individual case studies of particular lines. I have watched it happen often enough.

Now, if you've "restored OTP" by adding so much padding that the trains are no longer fast enough for a lot of people, you'll lose a bunch of ridership, but on the whole, the ridership is usually STILL better than the ridership when the schedule was unreliable. Really.

Click to expand...

Cynical question: Has extra padding ever actually brought OTP back up?

Click to expand...

Good cynical question! Certainly never on freight-railroad-operated lines. I've seen it happen on urban rail systems; it tends not to be discussed much, but you can spot it by the newspaper articles full of commuters bitching about longer runtimes.

Dunno about passenger-operator-owned intercity lines, because I haven't found any examples. There aren't that many examples of passenger-owned intercity lines in the US & Canada, and I've found very few examples anywhere of intercity operators "just adding padding" on lines they own.

Yeah, then why don't we have South of the Lake built yet? Why don't we have exclusive passenger tracks on the Empire Corridor? Heck, why hasn't the Point Defiance Bypass started construction yet?

Click to expand...

Because you guys haven't been spending the past several decades investing hundreds of millions of dollars into the rail corridors like California?

Click to expand...

Well, true for NY. But Illinois *has*, and so has Washington State. Illinois's money seems to have been largely burned on the St. Louis-Chicago line with little to show for it (I think we agree on how poorly this has been invested: money given to UP for track owned by UP, largely to raise the top speeds on the fastest bits). Washington, meanwhile, has been dealing with seemingly endless NIMBYs.

Politically, I think it is in large part dependent on commuter service;

Click to expand...

You may be right; the part of the Cascades route which got improved most promptly is the part with Sounder on it.

And the priority method of getting that needs to be state-owned track, because most of the Class Is have repeatedly demonstrated bad faith towards Amtrak in dispatching and/or maintenance, and so bluntly can't be trusted with track ownership. (NS has behaved itself very well since it lost the court case with NC over the ownership of the NCRR, but the very fact that that's the stick which changed their behavior should get you thinking.)

Click to expand...

It's not so much bad faith I think as that they've underestimated the amount of track that they need for their own freights (seriously, BNSF is less than thrilled over how screwed up the EB route is right now)

Click to expand...

You're giving CN, CP, and CSX far too much credit. BNSF seems to do its best with the dispatching, and has been rather apologetic. And right now, UP does its best with dispatching too. And I don't see anything to complain about from NS in the dispatching arena; I can't really complain about their "if you want the Wolverines to run on time, buy the track" blackmail a few years ago, because they did sell the track.
But CN was behaving really, really badly a few years ago (under Hunter Harrison) -- and was completely unapologetic, and issued blatantly dishonest "throw everything at the wall and see what sticks" excuses -- and CP is behaving really, really badly now (under Hunter Harrison).

And CSX dropped its maintenance standards to the point where it had excessive numbers of derailments, repeatedly, until there was a Congressional investigation. I suppose you could call this "underestimating the amount of track needed", but it's egregious. Since they fixed that, they're *still* not running the trains on time... but they seem more open to selling the lines to the passenger operator, which is good.

Wick Moorman of NS proposed a few years ago that the passenger operators should own all the lines and the freight companies should run on trackage rights. This would actually simplify things a lot for the freight railroads and probably make them a lot more profitable. The public subsidy to the freights would increase, but it would be in a rather indirect form: they wouldn't be paying the full cost of the added maintenance needed for tracks which carry heavy freight train. But it would also make the dispatching of the passenger trains a *lot* better.

and that you'll always have some degree of interference with mismatched running speeds; I have a hunch that OTP is directly related to the number of train meets and overtakes in a route; this would naturally result in terrible OTP for the long distance routes.

Click to expand...

There are places where this is probably true, particularly on single-tracked lines.

I do not believe that one can validly make such excuses for CSX on its fully-double-tracked-all-the-way-with-extra-sidings NY Central mainline, with widely spaced stations, where the number of passenger trains has been unchanged for 30 years. They're just not fulfilling their contractual obligations, and there's really nothing more to it. There's plenty of padding east of Schenectady and at Cleveland. You could say that they have trouble due to unexpectedly high passenger loads from Buffalo east, but they still don't deliver the trains from Buffalo to Cleveland on time...

Much worse, CP quite explicitly took money for track improvements intended for passenger train service on the Ethan Allen/Adirondack route from the government; the result has been substantially worse on-time-performance on that segment, which is totally unacceptable and indicates deliberate dispatching abuse.

Electrification (with 80mph at 79mph sections for ease of modeling, the impact is about 0.44 seconds per mile). It's probably a bit overoptimistic (and most likely overly pessimistic for the DMU), but it would be just under 5 minutes of time saving with no other improvements. Last bit of spamming that I'll be doing, just wanted to have a clear showing of the potential of electrification in combination with the other two showings.

Paulus,

What would you need to generate a similar file for the RF&P between Richmond Staples Mill and either Alexandria or Washington Union Station? I know that the line has a 70 MPH speed limit, but the prospect of knocking significant time off travel between Richmond and Washington simply by changing out equipment which can recover from curve slowdowns better (not to mention reducing the impact of adding another stop or two along the route) is a bit stunning. Especially since VA is going to want to buy its own equipment sooner or later as it is (for multiple reasons, not the least of which is going to be cutting the "rental costs" of Amfleets), DMUs might be something we'd want to look into out here (even if a toaster needs to get added at WAS, as long as the equipment can take NEC speeds even if they can't do it on their own...).

I've been using employee timetables (Metrolink and BNSF) for it. It's pretty much just a spreadsheet kitbashed together in Numbers and then copied over to Plot.ly to make the graphs. The hard part was finding the information for the trains, aside from that it's just time consuming as all get out. There's various degrees of fudginess involved of course (timetables go to tenths of a mile, I've been working with hundredths for accel/decel; timetable miles aren't necessarily 5,280 feet, etc.).

Honestly, if you assemble such a diagram for Richmond-DC, I expect you're going to discover that the trains could already be going a lot faster and are delayed by CSX freight prioritization.

I'm not sure what capabilites Plot.ly has for options, but with just 4 lines on your graph it's getting pretty muddled in areas. I took re-plotted your graph using Illustrator, I think its a bit easier to read; however multipul data streams on a coordinate graph is a difficult thing to make easy to read in general.



I also think it would be more useful if the chosen plotted trains were ones that could feasably make the jump over here. The Velaro model by Siemens (you already have) is a major canidate for state-side use; I would consider looking at the latest TGVs and Shinkansens as well.

peter

neroden said:

Honestly, if you assemble such a diagram for Richmond-DC, I expect you're going to discover that the trains could already be going a lot faster and are delayed by CSX freight prioritization.

Click to expand...

Oh, I know that. I've been on the Meteor when it's gotten into RVR 15 minutes early. If you got a good, clear run WAS-RVR and the passengers all had their act together at ALX you could probably drop 15-20 minutes off the Meteor's schedule in particular. Part of it is freight prioritization and/or capacity issues; part of it is also padding getting into WAS NB.

I'm not sure what a speed limit bump from 70 to 79 would do...with the exception of about a mile at Quantico Creek/Quantico station and half a mile at the Brooke station, you basically have 70 MPH all the way from ALX to the north side of FBG. South of FBG, the only slowdowns I see are at Doswell (for the diamond) and Ashland (for the run through the middle of town). It is quite possible that the curves can take something a little over 70, but I'm not sure what the limits would be.

neroden said:

Honestly, if you assemble such a diagram for Richmond-DC, I expect you're going to discover that the trains could already be going a lot faster and are delayed by CSX freight prioritization.

Click to expand...

It looks more like it's simply a function of longer trip runs adding more padding (which is to be expected if you're doing it on a percentage basis).

PerRock said:

I'm not sure what capabilites Plot.ly has for options, but with just 4 lines on your graph it's getting pretty muddled in areas. I took re-plotted your graph using Illustrator, I think its a bit easier to read; however multipul data streams on a coordinate graph is a difficult thing to make easy to read in general.

I also think it would be more useful if the chosen plotted trains were ones that could feasably make the jump over here. The Velaro model by Siemens (you already have) is a major canidate for state-side use; I would consider looking at the latest TGVs and Shinkansens as well.

peter

Click to expand...

Thanks, I'll see if I can tweak on Plot.ly and check out using Illustrator. The trains are selected more for "I can get the information" than anything else; performance characteristics can always be specified in accordance, it just may mean a less efficient train design is all.

Anderson said:

I'm not sure what a speed limit bump from 70 to 79 would do...with the exception of about a mile at Quantico Creek/Quantico station and half a mile at the Brooke station, you basically have 70 MPH all the way from ALX to the north side of FBG. South of FBG, the only slowdowns I see are at Doswell (for the diamond) and Ashland (for the run through the middle of town). It is quite possible that the curves can take something a little over 70, but I'm not sure what the limits would be.

Click to expand...

With DMUs, about 4 minutes on that 37 mile section of 70mph running south of Fredericksburg.