All Things Empire Builder

The westbound Empire Builder (train 7) is scheduled to serve Everett at 0838, seven minutes after the morning Cascades (train 510) to Vancouver. This seems like a missed opportunity to provide a convenient connection, as the next train is not until 1942.

Would it not be possible to run train 7 twenty minutes earlier west of Spokane, perhaps by putting locos at both ends and running push-pull west of Spokane? This would allow it to arrive at Everett at 0818, which would be a tight but achieveable connection onto the 0831 to Vancouver. Everett is a small enough station that it should be possible to make a 13-minute connection, even if the inbound Builder is a few minutes late.

Or is the present timetable etched in stone? I know that the Cascade Tunnel is a bottleneck. In order to optimise the use of this tunnel, BNSF presumably schedules its trains very precisely, and might send multiple trains in the same direction between each ventilation cycle. Moving Amtrak 20 minutes forward might be difficult to accommodate. Reinstating the electrification would potentially increase capacity, but would result in even more "platooning" than there is currently. There would be a once-in-a-generation opportunity to adjust Amtrak's timetable to make this and other connections as convenient as possible, but once this is done then we'll be stuck with the resulting timetable for years.

All of this assumes that Amtrak is actually able to run the Builder to schedule or at least within a few minutes of schedule. There is a fair amount of recovery time embedded in the timetable anyway, and the sectional running times seem fairly slack, so there's no obvious reason for the poor punctuality, unless freight operators are tired of slow passenger trains getting in the way of their 75 mph refuse trains and 90 mph intermodals. (To be fair, if a container train is late at a port then the containers might miss the ship and thus be delayed potentially by several days. If a coal train is late at the power station then the lights might go out. The dispatcher might consider either of these more important than passengers having to wait a few extra hours or Amtrak having to lay on taxis.)

serose wrote:The westbound Empire Builder (train 7) is scheduled to serve Everett at 0838, seven minutes after the morning Cascades (train 510) to Vancouver. This seems like a missed opportunity to provide a convenient connection, as the next train is not until 1942.

Would it not be possible to run train 7 twenty minutes earlier west of Spokane, perhaps by putting locos at both ends and running push-pull west of Spokane? This would allow it to arrive at Everett at 0818, which would be a tight but achieveable connection onto the 0831 to Vancouver. Everett is a small enough station that it should be possible to make a 13-minute connection, even if the inbound Builder is a few minutes late.

Or is the present timetable etched in stone? I know that the Cascade Tunnel is a bottleneck. In order to optimise the use of this tunnel, BNSF presumably schedules its trains very precisely, and might send multiple trains in the same direction between each ventilation cycle. Moving Amtrak 20 minutes forward might be difficult to accommodate. Reinstating the electrification would potentially increase capacity, but would result in even more "platooning" than there is currently. There would be a once-in-a-generation opportunity to adjust Amtrak's timetable to make this and other connections as convenient as possible, but once this is done then we'll be stuck with the resulting timetable for years.

All of this assumes that Amtrak is actually able to run the Builder to schedule or at least within a few minutes of schedule. There is a fair amount of recovery time embedded in the timetable anyway, and the sectional running times seem fairly slack, so there's no obvious reason for the poor punctuality, unless freight operators are tired of slow passenger trains getting in the way of their 75 mph refuse trains and 90 mph intermodals. (To be fair, if a container train is late at a port then the containers might miss the ship and thus be delayed potentially by several days. If a coal train is late at the power station then the lights might go out. The dispatcher might consider either of these more important than passengers having to wait a few extra hours or Amtrak having to lay on taxis.)

You'd be better served by delaying the Cascade service by 20 minutes, or (as now) assuming the passengers will catch the next Vancouver-bound train. While BNSF has good timekeeping, it's not perfect, and to guarantee a 15 minute connection with a train just ending a multi-thousand-mile journey along an exceptionally busy corridor, often subject to weather delays, is foolhardy at best.

Also, the 75 mph refuse and 90 mph intermodals you speak of... don't exist on this continent. 50-65 mph intermodal, 20-40 mph everything else would be more to the point.

I suggested only speeding up the Builder by 25 minutes west of Spokane as this seemed like it would have comparatively little effect on the timetable for the rest of the line. It would be possible to trim layovers elsewhere, and so maybe get to Everett 50 or 60 minutes sooner (at 0748 or 0738). This should (!) be a reliable connection to the morning Vancouver train, and might be a useful late-riser commuter train. Either option depends on the availability of paths through the Cascade Tunnel. I suppose, since Amtrak in theory has priority over BNSF, they can simply demand the path of the eastbound train in front of the Builder and then retimetable the Builder to keep this appointment with the tunnel.

With the forthcoming PRIIA review of the Empire Builder, and with the forthcoming introduction of PTC and potential line-speed improvements to 90 mph, presumably both Amtrak and BNSF are working on a fundamental timetable recast on the Northern Transcon. On any railroad, capacity is maximised by running all trains on the same track at the same speed, and having this speed as constant as possible. This means running refuse and coal trains at the same speed as passenger, mail and intermodal trains. If they're serious about capacity, they also need line-speed turnouts and long deceleration/acceleration tracks before stations and before any junctions onto lines with a slower line speed (or where the curve onto the new line can't be taken at line speed).

Railroads that try to have mixed passenger and freight traffic without off-line platform tracks at stations end up needing to allocate at least two, and often considerably more, train paths for each passenger train, even if freight and passenger trains travel at the same average speed. There's a reason that Gospel Oak to Barking can only have four freight and four passenger trains per hour per direction, when competent metro lines have 36 to 40 (passenger) trains per hour per direction. Maybe this is why UP need so much investment to run the Sunset Limited daily; building 110 mph turnouts and long deceleration tracks on each side of each station is expensive.

serose wrote:I suggested only speeding up the Builder by 25 minutes west of Spokane as this seemed like it would have comparatively little effect on the timetable for the rest of the line. It would be possible to trim layovers elsewhere, and so maybe get to Everett 50 or 60 minutes sooner (at 0748 or 0738). This should (!) be a reliable connection to the morning Vancouver train, and might be a useful late-riser commuter train. Either option depends on the availability of paths through the Cascade Tunnel. I suppose, since Amtrak in theory has priority over BNSF, they can simply demand the path of the eastbound train in front of the Builder and then retimetable the Builder to keep this appointment with the tunnel.

HA!

Suppose that Amtrak does demand such, and BNSF is forced to comply by regulators despite its best interests. First, BNSF officials will subtly, and without paperwork, mention to dispatchers that they shouldn't show such good timekeeping to the Builder. Perhaps it'll be a delay getting the Builder off Minnesota Commercial tracks and onto BNSF rails, perhaps a coal haul will accidentally be occupying the main through Northtown Yard and force Amtrak to use a lower-speed second track, perhaps the Devils Lake Sub will receive more oil trains and Amtrak will have more stop-and-proceeds. Next, BNSF will start to do the same across its system, affecting every train west of Chicago, excepting the Texas Eagle, the Lincoln service and the Missouri RiverRunner. At about that time, BNSF leaders will approach Amtrak leaders quietly and say, "We've been considering legal action to overturn RPSA '70 and PRIIA, since you know as well as we do that they're obsolete laws. Perhaps it won't work, but we have Warren Buffett's pocketbook behind us, and we can make life VERY difficult for you politically and economically."

Amtrak will work with BNSF within its bounds, but there is a limit on the demands it can place while a tenant.

serose wrote:With the forthcoming PRIIA review of the Empire Builder, and with the forthcoming introduction of PTC and potential line-speed improvements to 90 mph, presumably both Amtrak and BNSF are working on a fundamental timetable recast on the Northern Transcon. On any railroad, capacity is maximised by running all trains on the same track at the same speed, and having this speed as constant as possible. This means running refuse and coal trains at the same speed as passenger, mail and intermodal trains. If they're serious about capacity, they also need line-speed turnouts and long deceleration/acceleration tracks before stations and before any junctions onto lines with a slower line speed (or where the curve onto the new line can't be taken at line speed).

Railroads that try to have mixed passenger and freight traffic without off-line platform tracks at stations end up needing to allocate at least two, and often considerably more, train paths for each passenger train, even if freight and passenger trains travel at the same average speed. There's a reason that Gospel Oak to Barking can only have four freight and four passenger trains per hour per direction, when competent metro lines have 36 to 40 (passenger) trains per hour per direction. Maybe this is why UP need so much investment to run the Sunset Limited daily; building 110 mph turnouts and long deceleration tracks on each side of each station is expensive.

I do hope that Amtrak and BNSF (and CP Rail) will consider higher-speed running once PTC is installed. Obviously I can't say for certain, but the BNSF Northern Transcon ought to be good for 90 excepting the signaling, and CP Rail's ex-Milwaukee Road track is probably the same. Station sidings might help relieve some pressure on single-track segments in particular, though the major stations (CHI, MSP, MKE, SEA) tend to have them already and the minor ones tend to keep station stops to a few minutes.

Perhaps it'll be a delay getting the Builder off Minnesota Commercial tracks and onto BNSF rails, perhaps a coal haul will accidentally be occupying the main through Northtown Yard and force Amtrak to use a lower-speed second track, perhaps the Devils Lake Sub will receive more oil trains and Amtrak will have more stop-and-proceeds.

This would be BNSF cutting off its nose just to spite Amtrak's face. These extra oil and coal trains would delay intermodals as well as Amtrak. While there's some leeway in intermodal timings, there's not a lot as they have to fit in with customer requirements and ship schedules. While some westbound intermodals are carrying empty containers back to China, not all are, and shippers will not be happy if this stunt means delivery times are a week longer because they miss the boat.

As you say, this is a busy mixed-traffic railroad with (for the most part) at most one track per direction, and most trains aren't stopping at any given passenger station, so station sidings (including acceleration and deceleration tracks) are important however short the stop time is, just as they are on any real high-speed line. If there's no siding then the train has to decelerate to a stand and accelerate to line speed on the running line. These acceleration and deceleration rates are constrained by passengers' need for not falling out of bed or spilling their drinks. The freight train in rear must therefore be far enough behind to allow for this, and so must be further back than it would be from another freight train. Hence the need to book passenger trains into a double or treble path. The tighter the headway and the faster the passenger trains, the worse the impact.

Signalling for 90 mph trains is potentially a serious problem, especially for overnight trains. If we were discussing day trains you could just fit better engines and brakes and use the same block lengths, but passengers in top bunks tend to resent falling out of bed because the train brakes suddenly. So you need to resignal, and for any given type of signalling system the longer blocks reduce capacity yet further. Cab signals can help, since then you can have very short blocks and much finer control over train speeds. In principle this can give back some of the capacity lost by needing longer braking distances for 90 mph running.

I don't see adding sidings at every passenger station ever happening, due to cost and the fact that it might be physically impossible at some locations.

mtuandrew wrote:Also, the 75 mph refuse and 90 mph intermodals you speak of... don't exist on this continent. 50-65 mph intermodal, 20-40 mph everything else would be more to the point.

Don't UP run some of their PIG trains at 80 mph?

When I was working for Sperry along the BNSF Southern Transcon, the top speed for any freight was 79 mph, but that was only with certain approved equipment, basically single-level pig trains. Double stacks were limited to 60 I believe, and everything else was 45. I think, it has been awhile. The reasoning behind the lower freight speeds had to do with wind resistance (fuel usage and the law of diminishing returns) and the railcars ability to stay on the tracks. Ever see a railcar with a truck that is hunting at 40? It is a scary sight! Most North American freight cars are not designed with speed in mind, more focus is given to weight bearance and longevity.

In the short term, it would probably be easier to rejigger the Cascades timeing than the EB's, but over a longer period of time, there are certain capacity improvements that could, and will, be made to speed up the EB as well. PTC has some promise, but I am not sure it will be the magic bullet that it has been made out to be.

With conventional signalling, you usually have to leave a gap between trains on the same track in the same direction of at least 1.5 times the longest braking distance of any train on the line. This limits capacity, and also to some extent line speed. With cab signals you can have much shorter blocks, and also fewer constraints on where the block markers can be than on where signals can be. This allows headways to be reduced in two ways:

• the wasted block between two trains is a smaller fraction of the braking distance, and
• since braking distances for trains have to be rounded up to a multiple of the block length, the amount of space allocated can more closely match the braking distance.

Some forms of PTC include cab signals, or even moving-block (where the effective block length might be a few metres or less); such forms of PTC increase line capacity. Other forms may not increase capacity, but may allow higher speed through providing an assurance that trains will stop before any collision. Such an assurance might also allow buff-strength requirements to be reduced and so allow trains to be lighter (allowing faster acceleration and deceleration, and also higher speeds through curves).

Delaying the Cascades would mean the first train from Seattle to Vancouver wouldn't arrive until after noon, which might not be popular with Canada, nor might the changes to immigration staff shift times at Vancouver station. It would also lose the Seattle to Everett reverse-commute market (since it wouldn't get to Everett until after 0900). These trade-offs may or may not be worth it to add a connection from the EB without asking for a timetable change from BNSF.

Another way of solving the problem is to throw money and concrete at it. From the AAR's point of view, it'd be best to simply quad-track the entire Amtrak network and to electrify the pair of tracks to be used by Amtrak and express freight. Unfortunately, in the current economic and political climate it's unlikely that Congress would vote for that, and the AAR and its members won't pay unless mandated. More targeted construction might help; since the Cascade Tunnel is a bottleneck it might be worth doubling it, or at least reopening the old tunnel. Re-electrifying would remove the delays caused by the ventilation system, at a cost of adding delays to affix and remove an electric loco at either end, or of needing a fleet of bi-mode locos to run all trains passing through the tunnel. It might be worth considering bi-mode locos and a wider electrification programme in any case now that oil is as expensive as it is.

serose wrote:With conventional signalling, you usually have to leave a gap between trains on the same track in the same direction of at least 1.5 times the longest braking distance of any train on the line. This limits capacity, and also to some extent line speed. With cab signals you can have much shorter blocks, and also fewer constraints on where the block markers can be than on where signals can be. This allows headways to be reduced in two ways:

• the wasted block between two trains is a smaller fraction of the braking distance, and
• since braking distances for trains have to be rounded up to a multiple of the block length, the amount of space allocated can more closely match the braking distance.

Some forms of PTC include cab signals, or even moving-block (where the effective block length might be a few metres or less); such forms of PTC increase line capacity. Other forms may not increase capacity, but may allow higher speed through providing an assurance that trains will stop before any collision. Such an assurance might also allow buff-strength requirements to be reduced and so allow trains to be lighter (allowing faster acceleration and deceleration, and also higher speeds through curves).

Delaying the Cascades would mean the first train from Seattle to Vancouver wouldn't arrive until after noon, which might not be popular with Canada, nor might the changes to immigration staff shift times at Vancouver station. It would also lose the Seattle to Everett reverse-commute market (since it wouldn't get to Everett until after 0900). These trade-offs may or may not be worth it to add a connection from the EB without asking for a timetable change from BNSF.

Another way of solving the problem is to throw money and concrete at it. From the AAR's point of view, it'd be best to simply quad-track the entire Amtrak network and to electrify the pair of tracks to be used by Amtrak and express freight. Unfortunately, in the current economic and political climate it's unlikely that Congress would vote for that, and the AAR and its members won't pay unless mandated. More targeted construction might help; since the Cascade Tunnel is a bottleneck it might be worth doubling it, or at least reopening the old tunnel. Re-electrifying would remove the delays caused by the ventilation system, at a cost of adding delays to affix and remove an electric loco at either end, or of needing a fleet of bi-mode locos to run all trains passing through the tunnel. It might be worth considering bi-mode locos and a wider electrification programme in any case now that oil is as expensive as it is.

I do think that PTC, done correctly, can increase line capacity. BNSF and the entire Class 1 industry has been working on its own PTC system called ETMS, if you are interested in reading up on the subject - Amtrak would use this system when on freight company lines.

From the AAR's point of view, Amtrak is a nuisance that should be confined to lines that it owns, and annulled on other railroads. In fact, the AAR recently brought suit against Amtrak (one of its members) for excessive demands on its host lines. Contrast this with European and British models, where the passenger authority really is the power broker!

In regards to the Cascade Tunnel(s), Amtrak really ought to work with BNSF to get this issue resolved. The ex-GN line is the fastest current line between Seattle and Spokane by a good margin, and is essential to BNSF's Northern Transcon. However, there is another option - recently, BNSF investigated the reopening of the ex-Milwaukee Road line between Ellensburg, WA and Lind, WA as a cutoff of their ex-NP line which dips south to Pasco, WA. Such a line would give BNSF another direct line between Seattle and Spokane, which opens more space for Amtrak on either line. Perhaps the Feds and State can work out a deal with BNSF?

I can remember being on the morning Seattle to Vancouver train and the Empire Builder had yet to "clear" Everett. The Empire Builder was running ahead of schedule. And there were tranfers from the Empire Buider that got on the train to Vancouver, ironically. I suspect some sort of onboard pre-arrangements were made for said passengers on the Empire Builder (knowing they were going to get into Everett before the Amtrak Cascade train). So it's nice when #7 runs early and an make this connection at Everett, but I would imagine more times than not it doesn't make the connection, largely because passengers are not allowed to "book the connection."

At least under the present timetable, it doesn't help much if the Empire Builder is early at Everett. It's not scheduled to leave until 0838, and the Cascades is scheduled to leave at 0831. Since the Builder can't leave early (there's no 'D' or 'L' in the timetable), the Cascades has to have the platform line first so it can leave on time.

Everett has two platform tracks, one used by Sounder trains and one by Amtrak, and one non-platform bypass track. As far as I can tell from Google Maps, the only crossover onto the Sounder platform is at the Seattle end, so the only way an early Builder can have a platform is to go beyond the station and reverse into the Sounder platform. If the Sounder platform isn't available (e.g. on Saturdays and Sundays), the Builder would have to wait on the non-platform bypass track or outside the station, neither of which would endear Amtrak to BNSF, and neither of which would be safe for the passengers.

The point of making the Empire Builder earlier, or the Cascades later, is that it would then be the Builder which gets priority on the platform track in this sort of situation, and since the connections off the Builder seem to be tighter and more important than the connections off the Cascades it would seem sensible to reverse the order in which they serve Everett.

On the subject of ETMS, from what I can tell it does seem a very sophisticated system; indeed, I'm surprised BNSF aren't just using ETMS as the main signalling system, and dispensing with wayside signals altogeher. It'd save a fortune on signal maintenance. The lack of balises or beacons to indicate position is a bit worrying though. GPS is all very well, but it doesn't work well in tunnels or multi-track areas and simply stopping the train if it gets confused about where it is tends to delay the service.

I suppose PTC has to be sophisticated since there's a lot of variation in the performance of trains (even just freight trains). They can't simply precalculate braking curves and install overspeed sensors at the appropriate distance from signals (as the British did when they installed TPWS after Ladbroke Grove), since what is appropriate for one sort of train may be ridiculously long for another. Enforcing speed limits and work zones is relatively simple by comparison, as these are clear-cut. But I'm not sure how it's going to avoid going through incorrectly set turnouts. If a turnout is set to no route at all, or to a route occupied by another train, presumably this will already be detected by track circuits. Otherwise, will the train be told in advance which routes and diversions are acceptable? Will the engineer be able to override it if, say, there's a derailment at Pasco and train 27 needs to be sent non-stop to Vancouver WA via Seattle?

I finally got all the photos up.

Seattle to Wenatchee:
http://acm.jhu.edu/~sthurmovik/Railpics ... nails.html

Whitefish to Havre:
http://acm.jhu.edu/~sthurmovik/Railpics ... acier.html

Havre to North Dakota border:
http://acm.jhu.edu/~sthurmovik/Railpics ... lains.html

Fargo to Minneapolis-St Paul:
http://acm.jhu.edu/~sthurmovik/Railpics ... nails.html

Chicago:
http://acm.jhu.edu/~sthurmovik/Railpics ... nails.html

Capitol Limited:
http://acm.jhu.edu/~sthurmovik/Railpics ... nails.html

I'm very glad to hear the trip was overall pleasurable and relaxing. Upon first starting out reading the story, I couldn't help but think it was going to be an "Amtrak is so horrible" rant. All in all, Amtrak is not full of angry, incompetent people as outside soucres may suggest. Generally the staff in passenger services and operations are very knowledgeable - though always with exceptions. Especially in transportation and operations, the knowledge of the "politics" of being the guest on a host railroad and how to simply get things done can make or break a day/train/schedule.

A couple of things I know you'd love to control but never will -

Private cars. You have to get over it. Picking up, carrying and setting out private cars on the line-of-road is part of Amtrak's revenue. It's not a boxcar or 50 boxcars. It is an extremely delicate car filled with valuables. Picking up the stupid car takes time and setting it out takes time - especially if the train crew that day in that crew district hasn't so much as pulled a pin or made a hook in a LONG time - which is very common. These aren't freight employees who bang cars around for a living. There's a lot more to think about than just cutting away and running. Given the choice, I assume you would rather allow it to take some extra time than have something go seriously awry.

And schedule. It's agreed opon between Amtrak and the host railroad. If departure from Portland, Seattle or Chicago changes or any adjustments between are made it has to make TWO Class 1 railroad's approval as well as fit into the terminal schedule in Chicago and Seattle. That is the reality. 15 minutes here or there might not be a problem. But hours sometimes throws things way off and can't be agreed upon. Delays are what you make of them. On single track, paper railroads it's tough not to get stung especially with a lot of freight trains out there. Amtrak trains are generally the shortest on a given stretch of railroad, will fit anywhere to be holed up and can make the quickest moves to get out of the way (or run, depending on the dispatcher). It's very difficult to pick at delays caused by the host railroad without having the dispatchers board in front of you and even then, still hard to determine if you're getting nailed out of spite or necessity.

As far as getting your reservation situation straightened out, it sounds like Amtrak could not have done better. As already mentioned, if it was an airline you'd be sleeping in the terminal with no food available or getting a hotel room out of pocket.

Authorities have confirmed that 27-year-old Jared Nilles of Fargo, whose body was found Monday along the railroad tracks between Glyndon and Hawley, Minn., was a passenger on an Amtrak train bound for Chicago, Clay County Sheriff Bill Bergquist said this morning.
The Forum newspaper account:
http://www.inforum.com/event/article/id ... /homepage/

That Empire Builder was exactly an hour late leaving Fargo at 03:13 Monday 19 March. Victim fell off the train 20 miles later, which would be about 03:30. That's flat country and (pretty) straight track, speed limit probably 70. So really quickly after he boarded, he disembarked.

Foulplay? Disoriented? Unfamiliar? I dunno. Obit notes victim Jared Nilles was a college grad, a writer, performer and athlete. He was a Fargo native and attended school -- and worked some projects -- in Minneapolis, so I'm betting he had ridden the Builder before.

And kudos to that freight conductor who had pretty good eyes to spy a body on the right-of-way in the pre-sunrise twilight at 07:00. He must eat a lot of carrots.