Silver Star Update

[Ryan]

For those of us that don't have the luxury of a two day train ride, perhaps you can elaborate on what you're trying to say.

 

[Seaboard92]

Being from Columbia I can vouch for the intermediate traffic better then most. I normally board in Camden because the Columbia station tends to be very crowded almost like organized chaos. Every time I have gone to CLB to watch the train pull in or to ride I have noticed a fairly substantial crowd board (30-40+) in the coaches. One needs to remember the market for the Columbia station isn't just the Columbia metro. It is Spartanburg, Greenville, Anderson, Rock Hill, Asheville, and Charlotte for passengers headed to Florida. Headed north its a more localized market but it still has about the same ridership going north. Columbia also picks up some of the low country market and midlands markets as well, but those passengers have the choice of the Silver Meteor, Silver Star, and Palmetto. And its not uncommon for a few rooms to be vacated or board at Columbia as well.

Note: This is my train from one of my stations so I do have a obvious bias towards it.

 

[WoodyinNYC]

I just totaled up the consist counts and then took 20% as shop + protect, which is the way it is done now anyway.

9 * 1.2 = 10.8, which means roughly 11 cars needed for the Cap to have 3 each for 3 consists.

I don't think protects are computed on a per train basis. They are counted based on the grossed up totals, and distributed as necessary.

When I was counting protect cars individually, I was specifically assigning one each to Miami, New Orleans, and Chicago, and two to New York (or one to New York and one to Boston). For the Cap to be single-level, I figured one really ought to add one at DC. This number doesn't scale according to the number of cars in use; it's just a baseline.

On protect cars in general, not necessarily in this instance but maybe so, I feel like there's an "insurance factor".

That is, if Amtrak has one train running to New Orleans, two to Miami, and two different trains to Chicago, it could need at least one protect car in Miami, New Orleans, Chicago, plus 5 or 6 in NYC. (I'm ignoring Savannah here because no diner unless the Palmetto gets extended to Florida.) Because the "insurance risk" is that one car could go out of service on any one of these trains.

But if more trains are added out of NYC so there's four trains to Chicago (a second Lake Shore and a Broadway Ltd.) and three to Miami (the Palmetto restored to Silver Palm status), the risk that two of all these NYC trains will have a car go out of service on the same day is much lower than the chance that any one will. So adding three trains out of NYC should not require three more protect cars, Two more should cover the risk nicely, maybe only one more protect car operating at a slightly higher risk tolerance.

So growing Amtrak service substantially should not require quite as many protect cars. If we're talking the Viewliner order, that would save about $2.3 million per car, not nothing.

With the risk spread over a larger number of cars and trains, the "insurance cost per car/per train" could be cut. Merely one more car or one more train won't help expand the insured base enuff to spread the risk, but six or eight or 10 more cars or more trains probably would not require a 20% allowance of protect cars.

The cure for what ails Amtrak is more Amtrak.

 

[Anderson]

Paulus had that only 6% of sleeper passengers go between NYP and MIA. I was afraid that such might imply that the other 94% of sleeper passengers only travel for a short couple of stops (and have no need for any meals), and was hoping to get some clarification on that point.

Selecting NYP and MIA is a selective subset of the ridership stats for the Silvers. Orlando is a busier station for both the Meteor and Star than Miami. The top two city pairs by revenue for the Star are NYP-ORL and NYP to Tampa. The top two city pairs by both revenue and ridership for the Meteor are NYP-ORL and WAS-ORL. The Star gets a fair amount of intrastate FL business with TPA-MIA and TPA-WPB as the top two city pairs by ridership.
According to the 2013 NARP stats sheets, the average trip distance for the Star in coach was 471 miles and 844 miles for sleeper class. So the average trip in coach for the Star is still a longer distance then the NEC from WAS to BOS (457 miles). For the Meteor, the average trip in coach was 567 miles and 925 miles for sleeper class. Longer average distances than the Star, but not by a huge amount.

The other issue is that in the case of both trains, looking only at NYP excludes the significant number of boardings at PHL and WAS (as well as lesser numbers at BAL, WIL, TRE, and EWR). What I've long wanted is some sort of consolidation of the WAS-north trips on the Silvers (basically, treat everyone who uses the train from "Washington or somewhere north" as a single station for purposes of the stats. You also might want to consolidate everything down in SFRTA territory into a single "stop" for this as well: MIA boards something like 90k/yr but your total is closer to 250k for all the South Florida stops. Granted, a decent number of those are going to Tampa or Orlando...but plenty of them are continuing onwards.

Edit: Something to recall, also, is that with the Star you can basically get an "extra" day on your vacation vis-a-vis the Meteor in Orlando (e.g. departure around 1200-1300 vs. departure around 1900), so I can definitely see folks heading back to New York (who'd get back in around noon much of the time regardless) opting for the Star instead.

 

[jis]

It is quite safe to say that no one directly boards any of the Silver Service trains (or Palmetto or Carollinian for that matter) at EWR, since they do not stop there.

OTOH, there is some amount of transfer traffic that most likely happens ex SEC, EWR, MET, NBK and PJC, but those would be counted anyway at Newark or Trenton.

 

[Anderson]

...right, I flipped EWR and NWK. Anyhow, my point is that the numbers for "Northeast to Florida" are suppressed when you look at single city pairs and that the aggregate of the seven (or so) NEC stops on the Silvers is going to be a lot higher than just NYP-ORL or NYP-MIA.

Based on NARP's stats for the Meteor, NEC-to-Florida is somewhere between 25% and 35% of the Meteor's overall traffic. That's a guess on the basis of traffic traveling over 1000 miles on the Meteor (WAS-MIA is 1164; you've got a few non-NEC pairs that could qualify between there and 1000 but it isn't a huge number). 23% of traffic is over 1000 miles, but you've got a lot of pairs in the 800-999 mile range that might or might not be non-NEC (offering another 10.7%) and WAS-JAX is under 800 miles (700-799 miles has another 6.9%). So 23% of traffic is almost assuredly NEC-Florida and some share of the next 17.6% is as well. That's really an over/under of 23% and 40.6%; my best guess is that it's slightly towards the higher end (32.5-35%).

The Star is a bit more of a mess; my best guess is somewhere between 12.5% and 15% of traffic is NEC-Florida, but it is worth noting that the Star is flooded with short-hop traffic (the top 7 city pairs are under 350 miles, though 8 and 9 are NYP-ORL and NYP-TPA) and has more passengers overall (if for no other reason than a lot of seats get sold on both ends).

 

[jis]

When talking of NEC - Florida traffic, one might also wish to club together all the Florida destinations' traffic from NEC together to get a rounded up number for NEC - Florida, rather than considering just Miami or just Orlando or Tampa.

I read somehwere that when you do such clumping of airline traffic between NEC and Florida, it comes out to be the largest airline market in all of United States.

 

[Anderson]

When talking of NEC - Florida traffic, one might also wish to club together all the Florida destinations' traffic from NEC together to get a rounded up number for NEC - Florida, rather than considering just Miami or just Orlando or Tampa.

I read somehwere that when you do such clumping of airline traffic between NEC and Florida, it comes out to be the largest airline market in all of United States.

That's about right. IIRC, LAX-SFO and LAX-LAS are the two biggest air markets...but you have a slew of airports on each end which produce something like 40 main pairs (e.g. LGA-MIA, JFK-MIA)...and something like 20 of those pairs are quite large. A few are smaller (e.g. JFK, LGA, and EWR might not all dump much traffic on WPB) but I seem to recall coming up with this when I was tallying ridership once before. From what I recall as well, if Amtrak grabbed 5% of the air-rail market you'd need something like 8-10 long daily trains to deal with the load.

 

[Thirdrail7]

I'm just thinking aloud but what do you number crunchers think about the concept of yield management in situations like this? In other words, do you think Amtrak is "pushing" local travel to the Star by utilizing revenue tools?

The reason I ask is I believe (and I'm open to correction) the reason the Meteor has more capacity and accommodates longer distances is it connects to most of the system. The Star doesn't connect to many services without a significant wait, the Crescent being a notable exception.

I'm wondering if they "save" seats for longer trips on the meteor in the same way they used to "save" on some of the regional train by pricing the nearby trains at competitive rates.

 

[neroden]

On protect cars in general, not necessarily in this instance but maybe so, I feel like there's an "insurance factor".

Yes, protect cars are insurance. That's why you need the same number whether you run one train between the two endpoints or 10 trains between the two endpoints.

On the other hand, shop count is really a percentage of the cars, because each car is in the shop a certain percentage of the time.

This is why I'd really like a separate broken-out accounting for protect cars, on the one hand, and shop count, on the other hand. I don't have a good handle on how often Amtrak cars *ought* to be in the shop (required federal inspections, routine maintenance, etc.)

 

[Anderson]

On protect cars in general, not necessarily in this instance but maybe so, I feel like there's an "insurance factor".

Yes, protect cars are insurance. That's why you need the same number whether you run one train between the two endpoints or 10 trains between the two endpoints.

On the other hand, shop count is really a percentage of the cars, because each car is in the shop a certain percentage of the time.

This is why I'd really like a separate broken-out accounting for protect cars, on the one hand, and shop count, on the other hand. I don't have a good handle on how often Amtrak cars *ought* to be in the shop (required federal inspections, routine maintenance, etc.)

They're insurance, but the number needed does slide up (slowly) as the number of cars in service rises (with enough cars in service you can imagine multiple bad orders in a day); you also get to a point where you can tinker with your schedules and have a spare set ready to cover "meltdown delays" and/or to enable you to tighten equipment turns if you have enough trains in rotation. To offer an example, if you had a dozen overnight trains based out of NYP with roughly the same consists you can increase the number of same-day turns (and/or reduce the time needed for them) if you've got a spare consist that can be put out there when an inbound train is exceedingly late (e.g. the "spare Builder" situation in CHI).

 

[jis]

As long as Chicago 14th St and Sunnyside are unable to turn a consist in about 3 hours when push comes to shove, the only if will be to keep a lot of additional rolling stock stabled as protection.

As an example on Indian Railways for the Kolkata - New Delhi overnight trains (Rajdhani and Duronto) which covers the 900+ miles in 16-17 hours, basically starting late afternoon arriving late morning the method used is as follows:

All consists are owned by the Kolkata end and the Kolkata end maintains enough protect to cobble together one additional consist in general and upto two in winter. So roughly speaking 3 trains are run using 9+1 consists normally, 9+2 in the winter. Incidentally that already is a huge amount of protect, way more than Amtrak can afford I suppose, considering that each consist is 18 cars to sometimes as many as 20 cars. But I am told that protect equipment when necessary can be cobbled together from shop equipment by delaying shop service or expediting completion of shop work. So protect and shop numbers are somewhat mixed up, and there never is actually the entire protect pool sitting around doing nothing.

The trains are run as if they are running on a 32 hour mission with a long stop half way through their mission. They are prepared and maintained as such. Which means they receive basic cleanup, safety inspection and prep in New Delhi, and any major work is done only at the Kolkata end. Only on rare occasions is any equipment change made at the New Delhi end. The turn time normally in New Delhi is something like 7 hours. In extreme cases they are able to turn the consist in as little as two hours, which means upto 5 hour delayed incoming can be handled with little impact on the departure.

Do cancellations due to lack of equipment happen? Yes they do, specially in winters when fog delays exceed 5 to 9 hours for multiple days in a row. But there is no allocation other than non-maintainable levels of protect that could sustain service through such disruptions. So no one bothers to even try.

Basically the New York - Chicago trains could be managed using such a pattern, which would reduce the need for a lot of single level protect equipment in Chicago. Afterall we don't park Superliner protect equipment in Denver and Albuquerque. The consists are maintained with a requirement that they must run all the way to the end without requiring consist changes.

Such a methodology would also eliminate the ping-ponging of defective equipment between Chicago and New york with no one taking the responsibility to fix it. Responsibility would lie at exactly one place to fix it.

 

[Thirdrail7]

On protect cars in general, not necessarily in this instance but maybe so, I feel like there's an "insurance factor".

Yes, protect cars are insurance. That's why you need the same number whether you run one train between the two endpoints or 10 trains between the two endpoints.

On the other hand, shop count is really a percentage of the cars, because each car is in the shop a certain percentage of the time.

This is why I'd really like a separate broken-out accounting for protect cars, on the one hand, and shop count, on the other hand. I don't have a good handle on how often Amtrak cars *ought* to be in the shop (required federal inspections, routine maintenance, etc.)

To save yourself time, unless there is a waiver for continuous maintenance, figure something has to come out of line every 92 days for inspections. Additionally, there are major inspections every 2 years requiring various components to be completely disassembled and serviced/replaced.

This was one of the things Amtrak attempted to do last winter. They wanted to get ahead of these inspections during (what they perceived as) the slow season. It looked good on paper.

As long as Chicago 14th St and Sunnyside are unable to turn a consist in about 3 hours when push comes to shove, the only if will be to keep a lot of additional rolling stock stabled as protection.

As an example on Indian Railways for the Kolkata - New Delhi overnight trains (Rajdhani and Duronto) which covers the 900+ miles in 16-17 hours, basically starting late afternoon arriving late morning the method used is as follows:

All consists are owned by the Kolkata end and the Kolkata end maintains enough protect to cobble together one additional consist in general and upto two in winter. So roughly speaking 3 trains are run using 9+1 consists normally, 9+2 in the winter. Incidentally that already is a huge amount of protect, way more than Amtrak can afford I suppose, considering that each consist is 18 cars to sometimes as many as 20 cars. But I am told that protect equipment when necessary can be cobbled together from shop equipment by delaying shop service or expediting completion of shop work. So protect and shop numbers are somewhat mixed up, and there never is actually the entire protect pool sitting around doing nothing.

The trains are run as if they are running on a 32 hour mission with a long stop half way through their mission. They are prepared and maintained as such. Which means they receive basic cleanup, safety inspection and prep in New Delhi, and any major work is done only at the Kolkata end. Only on rare occasions is any equipment change made at the New Delhi end. The turn time normally in New Delhi is something like 7 hours. In extreme cases they are able to turn the consist in as little as two hours, which means upto 5 hour delayed incoming can be handled with little impact on the departure.

Do cancellations due to lack of equipment happen? Yes they do, specially in winters when fog delays exceed 5 to 9 hours for multiple days in a row. But there is no allocation other than non-maintainable levels of protect that could sustain service through such disruptions. So no one bothers to even try.

Basically the New York - Chicago trains could be managed using such a pattern, which would reduce the need for a lot of single level protect equipment in Chicago. Afterall we don't park Superliner protect equipment in Denver and Albuquerque. The consists are maintained with a requirement that they must run all the way to the end without requiring consist changes.

Such a methodology would also eliminate the ping-ponging of defective equipment between Chicago and New york with no one taking the responsibility to fix it. Responsibility would lie at exactly one place to fix it.

Such a methodology would also require a great deal of labor. If you had enough manpower and a fully staffed back shop, you could turn a train with defect in probably an hour.

However, those types of level are long gone. I was talking to an old Superintendent and he remarked he used to be able to turn a train including cleaning, cutting an engine on one end while adding an engine on the other end, turning the seats vacuums, interior windows being wiped in 15 minutes. If he had to dump the toilets and service the bathrooms, it took him 30 minutes as long as the train was spotted properly. He also told me that he used to have 6 electricians servicing the Southern Crescent alone!

Now, he's lucky if he has six electricians in two locations (yard and station) on his busiest tour.

 

[jis]

That indeed is a severe problem, so would probably make all this unworkable in the environment we have, unfortunately.

It also brings up a point that has been made here before which is, the way to fix Amtrak is to have more Amtrak. When you are sharing a pool of resources like maintenance persons over many more trains it becomes easier to justify them sicne you can spread the cost around. So if there were more LD trains it would be easier to adequately staff turn around activities - counter-intuitive somewhat, but seems to be the case.

 

[neroden]

To save yourself time, unless there is a waiver for continuous maintenance, figure something has to come out of line every 92 days for inspections.

Thanks! But *how long* is it out for inspections? If the inspections take one day, that's 1/92 shop time. If they take 10 days, that's 10/92 shop time.

Additionally, there are major inspections every 2 years requiring various components to be completely disassembled and serviced/replaced.

Again, thanks! But the key question is *how long* do these take? If the entire car can be disassembled, serviced, and reassembled in a day, that's 1/730 shop time. If it takes 30 days, that's 30/730 shop time...

Such a methodology would also require a great deal of labor. If you had enough manpower and a fully staffed back shop, you could turn a train with defect in probably an hour.

However, those types of level are long gone. I was talking to an old Superintendent and he remarked he used to be able to turn a train including cleaning, cutting an engine on one end while adding an engine on the other end, turning the seats vacuums, interior windows being wiped in 15 minutes. If he had to dump the toilets and service the bathrooms, it took him 30 minutes as long as the train was spotted properly. He also told me that he used to have 6 electricians servicing the Southern Crescent alone!

Of course, you can still support that number of maintenance workers if you're running a train every 15 minutes (or more). If not, you can't.... Economies of scale strike again.

Now, he's lucky if he has six electricians in two locations (yard and station) on his busiest tour.

At some point, cutting staff means buying more cars, which may be a bad tradeoff financially. Has Amtrak passed that point?
Thinking about it, it seems like it would be ideal for fast service to pack the maintenance workers into a very few locations, with almost all trains going to those locations, and with the yard basically co-located with the station. You could almost do this at Chicago, if that place wasn't so dysfunctional, and eliminate servicing at the other end for all the trains which go there. I doubt there's even room to do it at NY & Sunnyside, though, which is the other logical location.

Amtrak has some long-term poor locations, though. Beech Grove is just in a no-good location at this point, and it's not going to get better -- I don't see any scenario under which Indianapolis becomes a center of many-trains-per-day. Hialeah is a pretty awful location too, given how few trains Amtrak sends to Florida and the poor prospects for Florida in general (it's all gonna be underwater). If Beech Grove facilities were in Chicago and Hialeah facilities were in New York, the situation would be better.

LA is a nice consolidated location, but it just isn't serving enough trains to get those economies of scale (despite the Surfliners).

 

[Anderson]

On inspections: The 92-day inspection takes, IIRC, one day (plus any time to get it to/from the inspection). There's also an annual inspection that takes several days (and that I think has to happen in Beech Grover), but you could probably pile a bunch of those up between 9/15-11/15 and 1/10-3/10 (roughly the slow points in Amtrak's calendar...note that if the shops can handle it, the reduced consists in the winter are a good way to enable this).

 

[neroden]

OK, if we call the annual inspection 7 days and the 92-day inspection 3 days due to moving cars to the right place (this seems pessimistic), that's.... 19 days a year, or about 5% spent out of service.

Now, I assume that there is other routine maintenance based on when things *actually* break.

To get to a 20% shop count, you'd have to have something break taking each car out of service for an average of one day a week. OK, I suppose that's probably plausible... still seems high.

 

[Anderson]

OK, if we call the annual inspection 7 days and the 92-day inspection 3 days due to moving cars to the right place (this seems pessimistic), that's.... 19 days a year, or about 5% spent out of service.

Now, I assume that there is other routine maintenance based on when things *actually* break.

To get to a 20% shop count, you'd have to have something break taking each car out of service for an average of one day a week. OK, I suppose that's probably plausible... still seems high.

Actually, I think it's 16 days instead of 19 (the annual inspection does, I believe, cover one of the quarterly ones).

I think the 20% shop/protect count (remember, protect equipment is also part of that) also has to do with the small size of some of the parts of the fleet (e.g. 25 single-level diners, 50 single-level sleepers, 20 Acela sets) and needing to keep equipment in multiple places (NYP, CHI, MIA, and I think NOL for the singe-level stuff; WAS, NYP, and BOS for the Acelas). From what I can recall, you can slowly move that back to about 15% with a scaled-up fleet. Remember, with the dining cars, if you have to keep three as protect (NYP, MIA, CHI) that's about 12% of the diner fleet as protect equipment right there.

 

[Anderson]

Going with 16 days/year for stuff being out of service for inspections, with the diners do remember that with 25 diners you have 400 out-of-service days per year...which means that you need to account for two diners being out at once.

So, let's say that we need some sort of protect equipment four places (potentially): NYP, CHI, NOL, and MIA. WAS can arguably be sourced out of CHI if you wanted to convert the Cap, but other than that those four stations pretty much cover everything. So, let's scale equipment in a few tiers:
(1) 50 sleepers, 25 diners
(2) 75 sleepers, 26 diners
(3) 100 sleepers, 26 diners
(4) 200 sleepers, 51 diners

Tier 1 (present situation):
Sleepers (50): For maintenance you have 800 out-of-service days (assume three cars out at any given time). Protect here is 2 NYP, 1 CHI, 1 MIA, and 1 NOL allowing 41 cars in service reliably at any given moment. Distribution is:
-Meteor: 12 (4 sets, 3 each)
-Star: 8 (4 sets, 2 each)
-Crescent: 8 (4 sets, 2 each)
-LSL-NYP: 6 (3 sets, 2 each)
-LSL-BOS: 3 (3 sets, 1 each)
-Cardinal: 4 (2 sets, 2 each)
That totals to 41, which is right at your cap.
Diners (25): For maintenance you have 400 out-of-service days (assume two cars out at any given time). Protect equipment is 1 NYP, 1 CHI, 1 MIA. This gives you 20 cars reliably in service at any one time. Distribution is:
-Meteor: 4
-Star: 4
-Crescent: 4
-LSL-NYP: 3
-Cardinal: 2
That gives 17 and basically no room to add another diner; with the Heritage diners, you're actually coming up slightly short in some respects (hence the Star experiment) since there are only about 20 of those in use (though none are used on the Cardinal).

Tier 2 (post-Viewliner II order):
Sleepers (75): For maintenance you have 1200 out-of-service days (assume four cars at any given time). Protect here would be 3 NYP, 2 CHI, 2 MIA, 1 NOL, offering 63 cars reliably in service at one time. Distribution is:
-Meteor: 16 (4 sets, 4 each)
-Star: 12 (4 sets, 3 each)
-Crescent: 12 (4 sets, 3 each)
-LSL-NYP: 9 (3 sets, 3 each)
-LSL-BOS: 3 (3 sets, 1 each)
-Cardinal: 6 (3 sets, 2 each; assumes daily operation)
-Pennsylvanian: 3 (3 sets, 1 each)
-Shoreliner: 2 (2 sets, 1 each)
Diners (26): Basically the same as the above; you net one spare diner in the process (so 21 instead of 20)...but that gives you room to potentially add a diner somewhere. Distribution is:
-Meteor: 4
-Star: 4
-Crescent: 4
-LSL-NYP: 3
-Cardinal: 3
-Pennsylvanian: 3 (operated in lieu of a cafe car and with a different staffing situation)
That's 21, with 5 cars out as protect/shop.

Tier 3 (hypothetical additional sleeper order):
Sleepers (100): For maintenance you have 1600 out-of-service days (five cars). Protect here gets interesting, but I'm assuming 4 NYP, 3 CHI, 3 MIA, 1 NOL for 16 cars total. That allows 84 cars in service at once. Distribution is:
-Meteor: 20 (4 sets, 5 each)
-Star: 12 (4 sets, 3 each)
-Crescent: 16 (4 sets, 4 each)
-LSL-NYP: 12 (3 sets, 4 each)
-LSL-BOS: 3 (3 sets, 1 each)
-Cardinal: 9 (3 sets, 3 each)
-Pennsylvanian: 6 (3 sets, 2 each)
-Shoreliner: 2 (2 sets, 1 each)
That gives you 80 cars; there's room to add another one to the train of one's choosing and/or buffer protect allowances; you could, at this point, have "full cover" for a set of any train in New York and Chicago if you wanted to.
Diners (26): Same as above.


Tier 4 (and another hypothetical order):
200 sleepers/51 diners
Sleepers (200): For maintenance you have 3200 out-of-service days (I'm just going with 10 cars). Protect...I'd say 7 NYP, 6 CHI, 5 MIA, 2 NOL, 1 WAS. 31 cars out of service, 169 in service. Distribution would be as follows:
-Meteor: 32 (4 sets, 8 each)
-Star: 20 (4 sets, 5 each)
-Palm: 12 (4 sets, 3 each)
-Crescent: 20 (4 sets, 5 each)
-LSL-NYP: 15 (3 sets, 5 each)
-LSL-BOS: 3 (3 sets, 1 each)
-Cardinal: 12 (3 sets, 4 each)
-Broadway: 12 (3 sets, 4 each)
-Shoreliner: 4 (2 sets, 2 each)
-CONO: 15 (3 sets, 5 each)
-Cap: 21 (3 sets, 7 each)
That's 166 sleepers. Bear in mind tht in NYP and CHI, protect equipment can cover almost any one train running late (distributing those extra spares I haven't allocated would cover that, or you can put those on the Cap and take note of the fact that there are five trains coming into CHI and six into NYP, so you can use spares to shuffle sets in a pinch); in MIA it can allow any one train to be swapped for another train.
Diners (51): For maintenance you have 816 out-of-service days (3 cars). Protect...2 NYP, 2 MIA, 2 CHI. This gives you 42 diners, distribution as follows:
-Meteor: 8
-Star: 4
-Palm: 4
-Crescent: 4
-LSL-NYP: 3
-Cardinal: 3
-Broadway: 3
-CONO: 3
-Cap: 6
You've got four spares; I'd be inclined to park one in WAS, one in NOL, and the other two in NYP.

Excusing the wild-and-crazy allocations of equipment and operational issues in the last example, you basically have enough equipment that in order to get closer to 20% out-of-service equipment you're allocating masses of cars (basically entire extra trainsets) to each of your major hubs (NYP, CHI, MIA) and a modest amount elsewhere. The natural path (if you don't need to do this) is probably a few less spares (but maybe one or two more shop allocations).

 

[neroden]

I think the 20% shop/protect count (remember, protect equipment is also part of that) also has to do with the small size of some of the parts of the fleet

See, I'm absolutely sure of this. Larger fleets should require smaller shop & protect percentages, but it's hard to figure out how *much* smaller unless you break it down in detail. Economies of scale...

Anderson, thanks for trying to run some scenarios. It looks like the shop/protect percentage for a large class of cars should be considered to be 15% or less. 20% is suitable only for small classes of cars.

 

[Anderson]

Well, Amtrak has two other problems. One is the age of its fleet, which is probably adding a bit to shop counts on the Superliners (and do remember that there are five Superliner equipment types...six if you count the CCC as distinct from the others). The other is the need for a bunch of distributed spares (CHI, WAS/LOR, SFA, NOL, SAS, LAX, EMY, and SEA are all possible locations).

 

[jis]

They keep protect cars in SAS?

 

[Bob Dylan]

They keep protect cars in SAS?

Most of the time there's a Coach and a Sleeper in SAS as protect cars for the Sunset and Eagle jis! FTW sometimes has a P-42 and a Coach car sitting in the yard!

I personally think a spare P-42 is a better idea what with the breakdowns on the Single Engine Eagles being so common!!

 

[CHamilton]

NARP Chairman: Firsthand Report on Silver Star Sans Diner

 

[yarrow]

NARP Chairman: Firsthand Report on Silver Star Sans Diner

not surprising