Ideas for Additional "Night Owl" Train Service

[Anderson]

Well there is another private passenger lesson that could be applied in a way. Certain trains has dining cars but they didn't run the entire route. That would basically be like giving the Silver Star a Hamlet Washington dining car and removing it in both cities. Which saves you space in the crew dorm. But the switching of the consist makes it impractical. But a lot of trains back in the day did switching like that

The Crescent would be a particularly good candidate for this...an "enhanced cafe" along the lines of the LSL-Boston section would be sufficient south of Atlanta (traffic on the southern end being fairly coach-heavy as far as I can tell). A variation on this, based on what I saw in Queensland, would be to vary the crews on a given portion of a route based on the on-board traffic load (e.g. the California Zephyr probably does not need the same level of crewing between Grand Junction and Reno that it needs on either end of the route, while the Starlight is probably in the same boat between the Bay Area and Portland).

FWIW, I remember seeing the consists of some Florida trains (and indeed I think some of Southern's were the same way), and there would often be some Miami-Washington equipment. Especially given the interchange between the Cap and these trains, cutting off some cars at the same time as the locomotive gets pulled would not seem to be a stretch.

 

[A Voice]

Well there is another private passenger lesson that could be applied in a way. Certain trains has dining cars but they didn't run the entire route. That would basically be like giving the Silver Star a Hamlet Washington dining car and removing it in both cities. Which saves you space in the crew dorm. But the switching of the consist makes it impractical. But a lot of trains back in the day did switching like that

FWIW, I remember seeing the consists of some Florida trains (and indeed I think some of Southern's were the same way), and there would often be some Miami-Washington equipment. Especially given the interchange between the Cap and these trains, cutting off some cars at the same time as the locomotive gets pulled would not seem to be a stretch.

It is only another 225 miles up to New York (where equipment sets rotate), and the first car behind the locomotive you're changing is generally the baggage car. Lot of effort to switch cars out of the train, in a strong passenger rail market, to save merely a few hours.

 

[west point]

How much does it really add costs, though, to just let the "extra" cars tag along all the way to New Orleans? Anniston as a drop point is thinking outside the box, and Amtrak could do with more of that, but of course the idea is the additional capacity is only needed north of Atlanta.

The timetable distance between ATL <> NOL is listed as ~ 519 miles. PRIIA listed operating costs of a car is $4.00 + a mile.

A present day Crescent is usually 9 cars = 1 bag, 4 coaches, Diner, lounge 2 sleepers. If the ATL <> NOL segment was changed to 1 bag, 2 coaches, diner or lounge, 1 sleeper that would be 3 less cars every day.

3 cars x 519 x 2 directions x $4.00 = 9 = $12,456 saved per day

Now if Crescent goes NYP or WASH <> ATL 5 coaches and 4 sleepers = 12 cars then cutting them off at ATL saves another $12,456.

so regular consist saves in a 30 day month $$373,680. Proposed consist save double that in a month = $747.360. Those figures are only car operating costs. Operating savings per year ~ $ 4.2 - $8,4 M per year.

Additional costs in no special order. Since the cars are laying over and will be originating there may be 2 additional car knockers required for the FRA required daily originating inspections. ATL already has an unknown number for the 1000 mile brake check. Probably 2 switcher crew men. Then there probably would be some additional NS track charges ( wild guess not more than $5.000 ? ) Hotel power connections $1,000 month ? One time costs for additional spare parts, storage. & another vehicle.

Costs for 2 additional siding switches, tracks HEP connections, additional NS signaling, etc $1.4 - $2.0 M.

The problem with Anniston would be the additional car knockers and loss of savings of 200 miles a day. BHM looses 320 miles of savings.

As pointed out ATL <> NOL ridership is much lower part of Crescent's load. The Crescent has always been capacity limited ATL <> WASH even during SOU RR's operation. ATL <> NOL much less demand although BHM is adding more riders to train now both thru ATL and to NOL. Amtrak PRIIA stated more cars would be added ATL >? NOL whenever demand calls for extra seats. Good use for Horizons if they are ever removed from 1st line service.h

 

[Anderson]

Well there is another private passenger lesson that could be applied in a way. Certain trains has dining cars but they didn't run the entire route. That would basically be like giving the Silver Star a Hamlet Washington dining car and removing it in both cities. Which saves you space in the crew dorm. But the switching of the consist makes it impractical. But a lot of trains back in the day did switching like that

FWIW, I remember seeing the consists of some Florida trains (and indeed I think some of Southern's were the same way), and there would often be some Miami-Washington equipment. Especially given the interchange between the Cap and these trains, cutting off some cars at the same time as the locomotive gets pulled would not seem to be a stretch.

It is only another 225 miles up to New York (where equipment sets rotate), and the first car behind the locomotive you're changing is generally the baggage car. Lot of effort to switch cars out of the train, in a strong passenger rail market, to save merely a few hours.

The thing is, that 225 miles still totals about seven hours of runtime...and it has the same general impact as an Atlanta switch-out in spite of that saving about 22 hours on 522 miles.

Moreover, at least as of late the baggage car has been running on the back of some of those LD trains (as have the sleepers). While cutting the rear-most car may involve a more complex yard move to get it where it needs to go, doing so isn't horridly impractical (like cutting a mid-train coach would be) and it is possible to arrange a train in such a way that a cut-off coach and sleeper are next to one another.

I'll agree that doing this for a single car is potentially of limited value, but doing this with (say) a car each on the Meteor and Crescent gives you two cars. If you could find a logic to doing so on the Star as well (I'll happily grant that said logic may end up being a bit thin in Washington; Tampa would make the most sense but the timing doesn't work), or elsewhere on one of the single-level trains you'd save three cars...and that's an extra sleeper on the LSL, or indeed another cut-off car on the Meteor or Crescent. Two cars alone might give you room to experiment and/or run an extra less-than-daily/"demand-based" car on a train. Really, don't underestimate the value of preserving scarce capacity.

 

[Seaboard92]

If you were going to remove the dining car in sections of the Silver Meteor. You could run it MIA-SAV remove it there. And add one WAS-NYP and even that leg you might not need it as many people prefer to go to eat once they arrive somewhere. If you cut the Washington-NYP diner you would only need two cars.

 

[Anderson]

If you were going to remove the dining car in sections of the Silver Meteor. You could run it MIA-SAV remove it there. And add one WAS-NYP and even that leg you might not need it as many people prefer to go to eat once they arrive somewhere. If you cut the Washington-NYP diner you would only need two cars.

MIA-SAV might require some schedule tinkering NB, but I see what you're saying. On the northern end you'd probably need to put it on at RVR (or otherwise tweak the schedule) since you'd be pulling it at 1900 SB at Washington (and NB if the train is, say, an hour late things could get interesting). True, there are alternatives (one is to have some sort of meal package with an eatery in the station for WAS pax, the other is a Spirit of Queensland-type meal service) but that gets messy in some respects.

The Crescent is probably the best candidate for something like this: Have a diner WAS-ATL or WAS-ATN. The 1830 departure from WAS would allow for two full seatings (1830 and 2000/2015) while the timing into ATL/ATN would cover breakfast. Past ATL, traffic does drop off somewhat, and I think on this configuration you'd only need two dining cars. The next-best train for this would actually be a Cardinal on a slightly earlier schedule (presuming it had a diner): Run the diner NYP-Huntington and have some sort of "heavy Continental" service for sleeper pax on the way into Chicago. Depending on the timetable, doing the same thing with the LSL NYP-BUF would also be an option (and you've probably got a case for shuffling some cars at Buffalo...and as to the LSL's timetable? If they have to pad it out heading into NYP, they might as well get the benefit of optimizing equipment usage while they're at it...)

The real problem with doing this is that most routes are once a day. If there were, say, 4-5 trains on different routes, maintaining intermediate crew bases/turns would make sense. With that said, at some point we do get back to "drop off the crew, not the car" as being the preferable option (since you can at least in theory move a crew around to deal with something going haywire).

 

[jis]

As usual, I see that the trusty AU gang is busy exercising its brains trying to figure out how to cut service again

 

[Philly Amtrak Fan]

As usual, I see that the trusty AU gang is busy exercising its brains trying to figure out how to cut service again

Push another LD train under the bus so your train isn't the one (or one of the ones) that get(s) cut. I wish I was on this board in 1994 or 2004.

 

[west point]

Anderson The problem with these diner cut off cars is that it would cost more to cut off than continue on the train. Pre Amtrak the RRs could change out cars and add on cars using freight car employees to do the breaking of trains and combining them and doing the required brake test. Observed pre Amtrak trains usually do this in about 30 minutes with crews that usually had spare time anyway.

Now that is not possible with Amtrak needing at least 2 engine crewmen and 2 car knockers to cover the split. Then if timing is not correct you may have to double that number for a return train. 4 crewmen costs ? ? Not operating the diner at $4.00+ a mile can not save the costs of the additional Amtrak employees.

When figuring the costs of cut off cars at ATL it appeared that at least 2 - 4 cars cut off is a break even point for the Crescent's not taking cars <> NOL at 500 + miles.

 

[Ryan]

As usual, I see that the trusty AU gang is busy exercising its brains trying to figure out how to cut service again

Push another LD train under the bus so your train isn't the one (or one of the ones) that get(s) cut. I wish I was on this board in 1994 or 2004.

Or, here's a crazy idea - fight to keep all of the trains.

Insane, I know.

 

[Anderson]

Anderson The problem with these diner cut off cars is that it would cost more to cut off than continue on the train. Pre Amtrak the RRs could change out cars and add on cars using freight car employees to do the breaking of trains and combining them and doing the required brake test. Observed pre Amtrak trains usually do this in about 30 minutes with crews that usually had spare time anyway.

Now that is not possible with Amtrak needing at least 2 engine crewmen and 2 car knockers to cover the split. Then if timing is not correct you may have to double that number for a return train. 4 crewmen costs ? ? Not operating the diner at $4.00+ a mile can not save the costs of the additional Amtrak employees.

When figuring the costs of cut off cars at ATL it appeared that at least 2 - 4 cars cut off is a break even point for the Crescent's not taking cars <> NOL at 500 + miles.

Jis: It's not cutting service if the cars are effectively deadheading, and all the data I have seen on the Star, Meteor, and Crescent suggests that there is at least one car's worth of space running empty north of Washington (O/D at WAS for the Star is 49.5/train; for the Meteor it's 52.8 and for the Crescent it's 62.5/train). While I was poking around elsewhere, the most obvious candidate for cutting cars is still Washington, DC (where a locomotive has to be switched regardless), so cutting a sleeper and coach there as well really does make sense from the standpoint of "dead space" if doing so shakes enough cars loose to add something elsewhere (at least, presuming the trains each end up running with 3-4 sleepers) that's useful. The Meteor and Crescent are both timed such that this would work (the Meteor has a 12-13 hour overlap while the Crescent has a 9-hour overlap; interchanging the cars would give about 10-11 hours each, which is on par with what the Cap and LSL get in Chicago).

Washington is an odd case, however, and (again) that's why I raised the angle of tinkering with crew deployments rather than cutting out cars. There are trains where you can show that the relative pax load on one part of the route might justify full staffing but other parts clearly don't (I usually cite the Zephyr but I could probably cite the Builder instead...CHI-MSP has a very heavy load on that train but the crowd thins out a bit past MSP).

Edit: And to be clear, if $1m can either be saved or generated in incremental return through some changes like this (two well-utilized sleepers are probably worth close to $1m/yr)? That's $1m that can either go towards keeping a train around or towards equipment acquisition and so on.

 

[neroden]

With [correction: 20, not 22] added Viewliners in revenue service, each clearing roughly $750,000 a year, I'm hoping to see the Eastern trains as a group reducing their operating losses by some [correction: $15,000,000, not $16,500,000].

No point in killing off LD trains that are not even losing money.

The Viewliner trains, except the Cardinal, are already profitable before overhead allocation. This will make them more profitable.Cardinal needs to be daily to be profitable, though the Viewliners will help even while it's not daily. Amtrak needs to go ahead and figure out what needs to be done in capital upgrades to run a daily Cardinal with 3 consists (yes, Amtrak, it CAN be done)

 

[Anderson]

With [correction: 20, not 22] added Viewliners in revenue service, each clearing roughly $750,000 a year, I'm hoping to see the Eastern trains as a group reducing their operating losses by some [correction: $15,000,000, not $16,500,000].

No point in killing off LD trains that are not even losing money.

The Viewliner trains, except the Cardinal, are already profitable before overhead allocation. This will make them more profitable.Cardinal needs to be daily to be profitable, though the Viewliners will help even while it's not daily. Amtrak needs to go ahead and figure out what needs to be done in capital upgrades to run a daily Cardinal with 3 consists (yes, Amtrak, it CAN be done)

Per the "Boardman Chart", the Crescent also loses money (I don't know where the Star is now; at the time it showed a loss). As I think I've said before, though, I'm pretty sure that (1) it's in the black from NYP-ATL but loses money past there and (2) part of the loss is simply down to a lack of sleeper capacity on the north end. Bear in mind that the 28,640 sleeper pax the train has on two sleeping cars translates into 39.23 sleeper pax/train. Ignoring any lost demand due to the usual winter/spring cutbacks due to trackwork (in many years there's about a six-week period where the train truncates in Atlanta about 4x weekly) and realizing that some of one sleeper goes to OBS, even when you account for some of those spaces turning over at ATL/BHM, that seems to be pretty close to the best you can hope to do on a train that isn't flipping spaces like the Star seems to be.

 

[Carolina Special]

I have no difficulty accepting that shutting down any or all of LD trains would leave OH that would not completely go away and have to be absorbed by whatever is left. It does bug me to ignore allocations altogether in evaluating LD: if the allocations are being done correctly, then those expenses belong to LD whether these are regular expenses or coming from corporate. (Looking at the 2016 Five Year Budget Plan, Exhibit 3-6).

For example, the credit card fees allocated to LD from corporate. If the LD routes go away, the company total credit card fees will go down proportionately, absent any volume discounts: this expense belongs with LD. I expect you could go down the expense column line by line and and find a similar story, assuming the Amtrak accounting department is doing their job. I'd like to think they are and those allocations are there for a reason.

I very much respect the work on the individual train revenue/expenses being done on this board, but I don't see how you can avoid looking at allocations. I do not want to see any trains shut down, especially the Cardinal. But Moorman and others are correct that LD loses money based on the financial reports: we can't ignore that.

 

[Anderson]

I have no difficulty accepting that shutting down any or all of LD trains would leave OH that would not completely go away and have to be absorbed by whatever is left. It does bug me to ignore allocations altogether in evaluating LD: if the allocations are being done correctly, then those expenses belong to LD whether these are regular expenses or coming from corporate. (Looking at the 2016 Five Year Budget Plan, Exhibit 3-6).

For example, the credit card fees allocated to LD from corporate. If the LD routes go away, the company total credit card fees will go down proportionately, absent any volume discounts: this expense belongs with LD. I expect you could go down the expense column line by line and and find a similar story, assuming the Amtrak accounting department is doing their job. I'd like to think they are and those allocations are there for a reason.

I very much respect the work on the individual train revenue/expenses being done on this board, but I don't see how you can avoid looking at allocations. I do not want to see any trains shut down, especially the Cardinal. But Moorman and others are correct that LD loses money based on the financial reports: we can't ignore that.

Oh, I accept that many of the LD trains lose money. What I (and some of the others here) primarily object to is the tendency of Amtrak to use the LD trains as a "dumping ground" for mostly-fixed overhead expenses which makes them look a lot worse than they probably should, and by doing so make it harder to add service. "This train will add 250,000 riders and directly add $5m to the operating deficit" is a much easier sell than "This train will add 250,000 riders and directly add $50m to the operating deficit": The former loss is small enough that connecting riders might well obliterate the loss, and the loss-per-rider is $20. The latter looks like a massive hole in the ground. By the same token, posting that on a current train makes it look like eliminating that train would save far more than it actually would.

Unfortunately, it's functionally impossible for most of us to separate out the "justifiable" expenses being allocated with the non-justifiable ones.

 

[Philly Amtrak Fan]

Rough guess on allocation of costs onto LD trains

Definitely onto the spectific train:

Equipment, fuel, labor

If only train using track, track usage fees

If only train serving station, costs to maintain station including labor and maintenance.

On trains which share equipment with other stations/trains, split the costs. Ex. the tracks/stations in Florida are the same for the SS and SM so the costs there get split, similar to the LSL and CL between CHI and CLE. This is the economies of scale argument.

Using the NEC is a bonus, you don't have to pay usage fees and you can go faster, cutting down on fuel and labor costs.

Trains which have a lot of overlap seem to do better financially while those with unique routes (many trains out west) do worse. This is why a BL revival and a second LSL would be relatively cheap to Amtrak since most of the infrastructure is already in place. You say you don't want to discontinue routes or part of routes that have no other service but there would be significant cost savings as opposed to killing a train that shares most of the infrastructure already.

 

[Anderson]

Well, and remember that the LSL is also sharing tracks with multiple non-LD trains from NYP-BUF while the Silvers are sharing with other trains all the way down into NC. Of course, the Starlight shares with other trains for almost the entire distance from SAC-LAX and Eugene-Seattle...

The counter is simply that yield factors on the Silvers and the LSL are higher...particularly on the Meteor. Taking the Meteor, the #3 ridership pair is NYP-RVR. Coach on that segment is priced at Northeast Regional prices...which often means that coach pax are paying almost as much as sleeper pax (and indeed, I can find days that a roomette is cheaper than a flexible ticket), though Amtrak seems to have generally resolved the "its cheaper to book past RVR/PTB" quirk [1]. That said, the RVR-NYP coach price ranges from $78 (saver) to $193 (flexible). RVR-MIA, by contrast, runs $114-313. More expensive, yes, but that's on 335 miles vs 1054 miles, so your yield per passenger mile is $0.233-0.576 on the northern segment versus $0.108-0.297 on the southern segment. The sleepers are a similar (if less propotionally dramatic) story: For roomettes, I can find $183-298 RVR-NYP (so a yield of $0.546-0.890/mile) versus $399-650 RVR-MIA (so $0.379-0.617/mile).

I'd note that something similar applies on the Crescent and the Palmetto, if I'm not sorely mistaken. The Silver Star is a surreal wreck in this respect (the RVR-NYP sleeper fares tend to be about on par with Regional Business Class and are usually far cheaper than the flex tickets), with northern-end coach fares providing a mounting share of the train's revenue (RVR-NYP is the #1 revenue pair for the Star while NYP-RGH is #3; by ridership, those two are #3 and #9, respectively).

[1] I think this was probably fixed when they started allowing local traffic on the NEC, since eventually somebody was going to figure out that the cheapest last-minute NYP-WAS ticket was actually a NYP-RMT ticket.

 

[dlagrua]

I don't know what types of car pays for itself, but one of the arguments on this thread is that some business travelers would prefer to spend the night in a sleeping car and arrive at their destination city in the morning, rather than flying in the night before and spending $300 (or whatever) on a hotel. If you remove the sleeping car from the equation, the target market really becomes leisure travelers, who would be more price-sensitive.

The solution would be the slumber coach for the single level routes, that offered 24 one person private rooms (w toilet and sink)and eight double rooms for a capacity of 40 at a price just a bit more than coach. Those sleepers had more capacity than the Superliners of today. You will never see this type of car again because of the two steps up to the top row rooms and the lack of ADA accommodations. It was a great idea for travelers on a budget that wanted privacy and their own single bed. While roomettes are nice I wish that we had this low cost sleeper option again. It must have been popular with business people. The only one that I ever walked into was in the Illinois Railroad Museum.

 

[CCC1007]

I don't know what types of car pays for itself, but one of the arguments on this thread is that some business travelers would prefer to spend the night in a sleeping car and arrive at their destination city in the morning, rather than flying in the night before and spending $300 (or whatever) on a hotel. If you remove the sleeping car from the equation, the target market really becomes leisure travelers, who would be more price-sensitive.

The solution would be the slumber coach for the single level routes, that offered 24 one person private rooms (w toilet and sink)and eight double rooms for a capacity of 40 at a price just a bit more than coach. Those sleepers had more capacity than the Superliners of today. You will never see this type of car again because of the two steps up to the top row rooms and the lack of ADA accommodations. It was a great idea for travelers on a budget that wanted privacy and their own single bed. While roomettes are nice I wish that we had this low cost sleeper option again. It must have been popular with business people. The only one that I ever walked into was in the Illinois Railroad Museum.

Check the math again, a superliner has five bedrooms, thirteen publicly available roomettes and the family and accessible bedrooms for a total capacity of 42.

 

[WoodyinNYC]

The solution would be the slumber coach for the single level routes, that offered 24 one person private rooms (w toilet and sink)and eight double rooms for a capacity of 40 at a price just a bit more than coach. ... You will never see this type of car again because of the two steps up to the top row rooms and the lack of ADA accommodations.

Trying to picture this. You mean the one-person slumber rooms were stacked, in a lower row and a top row? So you couldn't stand up in them? Like sleeping inside an MRI scanning machine? Oy.

I'm thinking a single row of one-person slumber rooms, with sinks a la Viewliner IIs, toilets and showers down the hall, and stand-up-in ceiling heights.

Why wouldn't such slumber coaches meet ADA standards? Like, doors too narrow and not enuff space inside for a wheelchair? Then wouldn't, say, three or even two wider, accessible slumber rooms (in the space of four non-ADA slumber rooms), take care of ADA needs?

Guess that my idea is roomettes for single travelers, and wouldn't save enuff space to allow a good price cut from the regular roomettes.

 

[jis]

I had no problem standing up in Slumbercoach single rooms which were arranged in alternate upper and lower rooms. Look up information on Slumbercoaches. Amtrak ran them into the '90s.

 

[Seaboard92]

Here is the exterior of a Burlington Slumbercoach.

 

[Anderson]

Depending on the configuration you've got a few options:
(1) An all-roomette car. This is the simplest option (some of these were made in "the day"). I'm thinking the configuration would be 22 roomettes for sale, one for a crew member (or perhaps two attendants per three cars), and one for a pair of toilets (no shower). You might need to trade one more roomette off for some sort of ADA space, but 22 roomettes might well be doable.

(2) Slumbercoaches:
s1346.photobucket.com/user/VPayne1/media/USPat25561401945_zpsb4c24512.png.html
32 spaces, so you could arrange about a 25% discount and still come out ahead. Note that you can stand up in a slumbercoach; the configuration is just wonky.

 

[neroden]

Per the "Boardman Chart", the Crescent also loses money (I don't know where the Star is now; at the time it showed a loss).

Yeah, I've basically been updating that chart each year by the following algorithm: I took the difference between the direct-costs loss and the fully-allocated loss that year (took us a while to get the years lined up, but we did it, I think it was 2012) and used that as the "allocated costs" number.(I updated the allocations total the last time I got a good number for the total amount assigned to the long-distance line as a group, in 2014. It had gone up, so I increased all the allocated numbers proportionally at that time.)

If you subtract this allocated costs number out from the current "fully allocated losses", you can figure out how the trains are doing this year. If the allocations have been increasing (which they appear to have been) then I'm actually underestimating their profitability.

The only two ways I could be overestimating is if (a) allocated costs went down significantly since 2014, which is highly unlikely due to inflation alone, or (b) there's been a systematic reallocation of costs away from the LD trains to the NEC, and I'm pretty sure there hasn't been yet.

Anyway, it looks to me like the Star and Meteor are now both *highly* profitable, while the LSL and the Crescent are *slightly* profitable (the Crescent is just this side of breakeven, less than a million a year by my calculations).

I wouldn't have to do all this calcuation if Amtrak would just release the direct cost numbers like they're supposed to (mandated by PRIIA 2008) but they haven't. I would really like it if they would because I do think I'm underestimating the profitability, and I think Amtrak could show some really good numbers.

I dunno, maybe Amtrak has some obscure political reason for not admitting that the Eastern trains are profitable. Maybe it's because the Transcons are still actually unprofitable on a direct costs basis? It is possible that publishing the numbers would cause a movement to leave the profitable trains alone and kill the CZ, SWC, and SL. On the other hand given the political situation I think it's worse to let Congress mistakenly believe that the LSL is unprofitable, when it's really profitable.

If you want to know the major cost drivers in the allocations, it is NOT stuff which is truly proportional like credit card fees; it's stuff like IT which isn't proportional at all, and the costs of operating the backshop, which is only *partly* proportional and is apparently simply not accounted for in enough detail to know which parts are avoidable (but most of them aren't; the workers are on staff regardless of how many train services are running). The last half-decent breakdown of these allocated costs was in the 2014 annual report IIRC.

 

[neroden]

Well, and remember that the LSL is also sharing tracks with multiple non-LD trains from NYP-BUF while the Silvers are sharing with other trains all the way down into NC. Of course, the Starlight shares with other trains for almost the entire distance from SAC-LAX and Eugene-Seattle...

If you do what I did and subtract out a reasonable underestimate of the allocations, you find that the Starlight is now the best-performing Superliner train. It might be tipping over into direct-cost profits in 2017 or 2018. This is not coincidental. Sharing tracks with those other trains is a big deal.
I should also note that the Crescent has overlap with the Lynchburg train and the Carolinian/Piedmont.

Honestly, when I sort the trains by estimated direct-cost profits/losses? With two exceptions, it looks like a list of how much overlap they have with other train services:

Auto Train

Silver Star

Silver Meteor

Palmetto

LSL

Crescent

Coast Starlight

Cardinal

Empire Builder

City of New Orleans

Capitol Limited

Texas Eagle

Southwest Chief

California Zephyr

Sunset Limited

The exceptions are, of course, Auto Train and Empire Builder (significantly better than you'd expect if you predicted by amount of overlap). You could also argue about the relative ordering of the Cardinal (slightly better than you'd expect) and the Texas Eagle (slightly worse than you'd expect), but you see my point.

This shouldn't be surprising, of course.