Learning from the Keystone Corridor

I have to ask the question who’s side are you on?

Using the whole Subsidy per Passenger Mile statistic is pure BS. It is a made up stat to imply that running no train at all is the best financial position. If we the people who are trying to get more and better passenger rail keep putting inaccurate information out there, we should expect the results to be the same we’ve always had. To expect different is close enough to the definition of insanity.

The first part of the article was positive and then it just flipped…

” The investments made in 2005 and 2006 on catenary upgrades and track replacement were made in the context of an already high-quality line”
This statement is not true. it was formerly a high quality line but it’s conditions were deplorable. Stick rail made the ride feel like a vibrating chair. while the scheduled time may have been 2 hours, the train was NEVER on time before the upgrade. Moreover, the upgrade dates to 2003 when work started. Slowly over the period the ride quality improved as did OTP. I’d credit much of the ridership since 03 to the upgrade. The line still isn’t what it shoudl be but it’s a decent line now. Additionally, the subsidy figures are misleading. I’m assuming you’re using PennDOT numbers but these numbers are contract numbers (ie they represent 51% of the total revenues attributed to the line so if it’s $7.2 million, actual attributed revenue is $14.4 million). More to the point, some of the revenue is attributed to the NEC since many trips go to NYP or some other NEC destination. In all likelihood, the total revenue of the line from Harrisburg to NYP is much, much higher than is stated here. the state subsidy basically covers “the gap” as calculated by amtrak.

Since the point of operational profitability has come up, I heard that the Chicago – Twin Cities line would be operationally profitable with seven roundtrips a day making the journey in 5.5 hours, which is a small improvement over driving but a huge improvement over the Empire Builder (8-8.5 hours).

And that would still be diesel within a freight railroad, but mostly with separate track. The Midwest High-Speed Rail study also stated in the early 2000’s that the St. Louis & Detroit corridors would be operationally profitable.

Just because the rhetoric of profit and subsidy rate are used by anti-transit activists doesn’t mean that those factors should simply be tossed aside on a scarp heap. That’s a reactionary response and precisely where we go wrong in this country with transportation (and most everything else,) sacrificing functional discussions in favor of detached ideology-driven political tug-of-war.

Why is subsidy a relevant discussion point? Precisely because subsidy will be required for many, perhaps most, maybe even all services. Knowing which would require the least subsidy can be a great tool for determining investment priorities. If you have only a limited amount of funding, it makes sense to look at both popular *and* low subsidy routes as initial investments to build ridership.

It can also be relevant to transit systems seeking to stabilize their systems. Here in Portland, the subsidy per rider on the MAX light rail system is far less than the subsidy on the bus system. (In fact, depending on the numbers, some claim MAX actually makes a profit at the farebox.) When the goal is to build a total transit system, the ability to shift significant portions to revenue neutral or revenue positive modes is absolutely vital.

In short, while reducing subsidy is not the only or even the most important factor in establishing or improving transit and transportation, it is still a very useful and important one that should not be ignored simply because political opponents misuse it.

Cameron, Chicago-St. Louis service is already operationally profitable (barely) in good years. In bad years, nothing is operationally profitable except the Acela, which gets to piggyback on the breaking-even-at-best Regional.

It’s possible. The PRR used to run trains from Washington and Baltimore to the Main Line, joining the Broadway Limited going to Chicago. However, such a connection right now is low-priority – it’s much more important to upgrade the NEC to HSR specs, and then construct Philly-Pittsburgh HSR.

Don’t underestimate the value of those last few upgrades on the harrisburg-philly line. Closing those last three at-grade crossings will allow true max 110 mph speeds across the whole line. And closing those at grade crossings should also then allow max speeds to go up to 125 mpg assuming there aren’t more curves that need to be straightened out.

Eldondre: on the contrary, Philly-Pittsburgh is more more expensive than the NEC, since it requires tunneling under the Appalachians. The NEC requires a few short tunnels under suburbia, but the total tunnel length is overall much shorter.

Getting Philly-Pittsburgh to 5 hours will be a complete waste of money. Stop investing in legacy rail – learn from France and Japan, and either spend the money on full HSR, or don’t build anything.

alon, a five hour trip would not require tunneling under the Appalachians, a 2.5 hour trip would.
Adam-agreed, those last few pieces are important and what’s more, the east of harrisburg will likely stay mostly intact should real HSR ever be built while west of Harrisburg would require an entirely new ROW . There are a couple of curves east that should eventually be straightened. For the price it’s a worthwhile investment. I believe the 2004 study for three trains west of harrisburg at 5.5 horus was 110 million.

It’s gonna cost A LOT more before this gets done. Another billion to Harrisburg? I’m with Bruce McF here. It’s only half a million people in the metro. And I’m stumped to think why anyone not involved with the state government would want to go there. Just sayin.

Excerpts from a muddled article in the Central Penn Business Journal:

By Jim T. Ryan, 7/17/2009

Gov. Ed Rendell is seeking … (ARRA) money to plan for high-speed rail transportation, including about $1.4 billion for upgrades to the Amtrak Keystone Corridor East between Harrisburg and Philadelphia, said Erin Waters, a PennDOT spokeswoman.

The corridor could have train speeds upgraded to 125 mph following proposed removal of road crossings. It now has trains that run at 110 mph.

Eliminating the crossings would cost about $418 million, Waters said. The two-phase project, including upgrades to power supplies, would cost about $1 billion.

PennDOT also submitted plans … for a feasibility study to add high-speed rail to the Keystone Corridor West, which runs one train a day between Harrisburg and Pittsburgh. The study would explore upgrades to the Norfolk Southern line, addition of a parallel line and new locations for the high-speed rail.

Alon-It’s standard speed rail and even in countries with real high speed, they have cheaper standard speed alternatives. At five hours it will attract a lot more riders than it does currently, and less than if it were 2 hours.
AFAIK, the $1 bn dollars cited by PennDOT was a ballpark figure and included prepping the line for speeds of 150 mph AND new equipment. the Phase I request was $28 million for engineering and $750k to study alternatives to Pittsburgh, entirely within the realm of reason.

Ridership-that does include All riders on the line but the revenue does not. If Yonah has a question abou the revenue he can feel free to email me. I corrected the Inquirer as well who verified that APTA was using the PennDOT number which itself does not agree with Amtrak numbers. For all the complaints about harrisburg, the state subsidizes about $6.66 per passenger, the feds subsidized Cape Air in Lancaster to the tune of about $110 per passenger.

Which is why the phrase “at most” was used.

Fast Eddie’s sticky fingers said on September 29, 2009 at 09:42:
“And an extra $1 billion to go from 110 to 125mph? Do the math, on this 104 mile corridor going from 110 to 125mph can at most save 6 minutes.”

Yes, precisely, do the math. If its 104miles and the fastest schedule is 1:35, that’s an average trip speed of 64mph. If the 125mph upgrade can get the trip speed up to:
* 80mph will save 17 minutes
* 90mph will save 25 minutes
* 100mph will save 32 minutes

Obviously, the “125mph upgrade” will not be restricted to upgrading 110mph sections by 15mph, 90mph sections by 12mph, 30mph sections by 4mph and 15mph sections by 2mph, as your calculation of a 13% increase in average trip speed implies.

I think eldondre is right about the target timetable. I recall reading somewhere, but can’t find it now (my search skills are not so great), that the aim is to get the Keystone (East) down to 75 minutes. One hour 15 minutes Philly-Harrisburg would be 20 minutes less than now.

By the time the Keystone Corridor upgrades are finished, it will be almost like a new line. When Amtrak started on this project, it owned the tracks, so this was where it could do a demonstration project. And Pennsylvania has been a willing partner, putting in considerable resources of its own.

Actually, the other place where Amtrak owns the ROW has also seen considerable work, between Kalamazoo and the Indiana shore. Michigan put in some resources as well, closing grade crossings and building a highway overpass or two.

But now Michigan is broke. And the big end of the unfinished work on that route is in Chicago. Until the CREATE projects get that mess untangled, with the hope of cutting 15 or 20 minutes off the schedules of all the long-distance Amtrak trains out of the city, there’s little point in sinking much more money into the ROW to Kalamazoo and beyond to Battle Creek, Ann Arbor, and Detroit.

So Yonah is correct that the Keystone is unlike almost any other potential HSR corridor.

The data remains very mixed. I went to Amtrak’s latest monthly figures. For July, ridership on the Keystone service was down 3.4% from July 2008, compared to minus 9.6% for all state-supported and short distance trains. It outperformed the NEC spine Acelas and Regionals, down 8.4, and the long-distance trains down 7.7%. (The ONLY trains with a plus mark were the Lincoln service, St Louis-Chicago, +9.4%, and the Palmetto, up 0.9%. The Heartland Flyer was the worst, down 26.1%.)

A provocative post by Yonah and great discussion. But I’d be very careful before drawing sweeping conclusions about what the Keystone shows about the value of incremental improvements to 90 mph or even 110 or 125 mph service linking small metro areas vs true HSR between major metros. I still favor both.

the plan is detailed here
http://planthekeystone.com/assets/downloads/Keystone_Environmental_Assessment.pdf

As for midsized metros, were PA not caught with it’s pants down I assume they would have submitted something for Pittsburgh. OTOH, this line is part of the Pittsburgh route so it does go towards the goal of eventually linking Pitt with the east. They expect ridership to increase to 1.7 million with these upgrades (assuming no additional trains, which seems like a questionable function since 20 minute trip savings would surely free up equipment for more runs, particularly at rush hour). I’d be curious what kind of time savings tilt equipment would achieve. I’d imagine the holy grail here would be an average speed of 104 mph (60 minutes HAR to Philly). Prior to that, though, I hope they’ve lopped 2 hours off the western leg.

eldondre — Thanks for the great link. Good to see the detailed info. Hey, if they can get a 50% increase in passengers up to 1.7 million a year in just a few years, that’s pretty good for a mid-sized market. And of course, they did not assume additional trains, as you point out, not even a second run to Pittsburgh. One of these days.

Alon Levy on September 29, 2009 at 15:04:

“Bruce: Japan’s narrow-gauge trains still top at about 160 km/h, and usually don’t go above 130.”

A top of about 160mph is higher than a top of 79mph. When elaborated, your position seems to be that the US should make no investments in rail corridors to bring the top speed above 79mph in the coming decade, but should instead wait until after the first Express HSR corridor has been built in a region before starting to make any investments that make any higher than 79mph rail services available in that region.

“The problem with the Emerging HSR concept is that the investments done for it are exactly the types that don’t go well with national HSR networks. It involves grade crossing eliminations, some curve straightening, and a lot of double-tracking. There’s no electrification there – the idea that someone might want to run modern tilting trains on the line at lower speed and then join the HSR line at high speed just hasn’t occurred to the FRA.”

You have either confused yourself or are playing a dishonest rhetorical game. “There’s no electrification here” is true in the sense that there is no requirement for electrification. It does not mean there is no scope for electrification.

But its only if it electrification was ruled out that the argument makes any sense. If there are corridors that can generate operating ratios above 1.00 with 110mph diesel service, they can refund revenue bonds to finance electrification, which will in turn increase patronage due to reduced trip times and reduce operating costs, increasing future operating ratios. And of course, electrification work is among the projects that can be funded in the Federal HSR funding framework.

I personally advocate a strong downgrade in competitive applications if infrastructure is being funded that is incompatible with future electrification as an explicit policy, but painting the current policy as obstructing electrification would be an absurd misrepresentation.

“Also: the 125 mph upgrade won’t increase average speeds the way you say it will. The only parts for which the top speed is relevant are those where trains can attain their top speed. You can eliminate curves and other bottlenecks, raising the average speed, without ever raising the top speed.”

That was in response to someone who painted a project application that includes upgrade of 110mph and 90mph segments to 125mph as consisting of nothing but an upgrade to 125mph, and ignored the travel time savings offered by the components of the project in favor of a simplistic projection of the 13% increase in maximum speed from 110mph to 125mph.

And, obviously, where time elasticities are stable, eliminating lower speed bottlenecks increases the patronage benefit of upgrading to 125mph, since after the reduction of the bottleneck, the same time saving is a larger percentage reduction in travel time.

A top of about 160mph is higher than a top of 79mph

Not 160 mph – 160 km/h. And it’s mostly 130 km/h, which is 81 mph.

You have either confused yourself or are playing a dishonest rhetorical game. “There’s no electrification here” is true in the sense that there is no requirement for electrification. It does not mean there is no scope for electrification.

It means scarce investment dollars are spent on a diesel-based system, which will with few exceptions have operating ratios far below 1. If the Keystone and Empire Corridors can’t raise their operating ratios to 1, neither can most other lines. This ensures that not only will those lines be incompatible with HSR for the near future, but also they will force Amtrak to spend more money on a system that will remain incompatible with HSR.

Eldondre, Don’t forget the run from Philly to Pittsburgh is also toll road. So there is an extra costs factor in that equation.

To all…

The key would be to electrify the line, and continually work on upgrades.

At the same time if you wanted a new tunnel in places to make it better, go for it. But they should not stop spending for service now, because they’re building a better system for the future. That is how the future never comes.

That’s exactly right alon. In my mind, the HSR segment should be a non-stop Harrisburg-Pittsburgh segment with the old ROW continued to be served by “regular” trains. I don’t see an entirely new ROW east of Harrisburg though it may be worthwhile straightening some curves.

Again my compliments, Yonah, for provoking so many comments (43 and counting) and informative discussion with this post. But as I go over it once again, I’m still unsure. The PA statistics you cite give revenue as $7.2 million last year..

When I look at the Amtrak monthly report —
http://www.amtrak.com/pdf/0907monthly.pdf — I see $2,244,000 for July 2008 alone, times 12 crudely gets me almost $27 million revenue for an artificial 12-month year. (Everything is down for 2009, of course, deep into The Great Recession.)

Presumably Amtrak reports the revenue from the Keystone trains all the way to NYC, and counts the trips like Lancaster-Trenton and Philly-Newark, and not just those solely within the Keystone East Philly-Harrisburg Corridor.

If the total revenue is closer to what Amtrak indicates, then the value of the incremental improvements may be higher than the lower PA-only figure suggests.

Alon is insisting that upgrades to less than true HSR are only justified when the routes are electrified and the HSR trains can run on them before joining the HSR trackage.

Isn’t that what is happening here? Improvements on the Keystone East are helping to feed more traffic from Harrisburg, Lancaster, and the smaller stations onto the NEC.

In that case, when Amtrak spent $72..5 mil and PA matched that to get $145 mil in improvements to put more passengers on the Keystone-acting-as-regional trains up to NYC, it was a deal. Compared to the many billions it will cost to shave even 5 or 10 minutes from timetables on the NEC itself, another billion to cut 20 minutes from the Keystone feeder route will be a deal.

From the interchange called Pittsburgh to the one called Philadelphia, the PennsylvaniaTurnpike costs $21.60. (Save 4 cents if U use EZPass.)

“The revenue used here isn’t even the full revenue for the Harrisburg segment,” says eldondre. Well, in that PA report that Yonah cites, the section, pages 126 & 127, is headed “Commuter Rail Program.” I wasn’t thinking this Harrisburg-Philly we are talking about is exactly “commuter rail.”

Peer at a sort of footnote to the header above the schematic map on page 126: “Beginning 10/06, Keystone rail service expanded, and the Commonwealth began receiving credit for 51% of all Keystone riders, revenue, and passenger miles incurred west of Philadelphia.”

Does that almost inscrutable bureaucratese lingo mean that the $7.2 million “Annual Passenger Revenue” figure on page 127 is half the true total , and Amtrak gets credit for the other half, or what the heck?

As a retired journalist, I often scold younger reporters for garbling stores about rail and transit. But my sympathies go to any journalist on deadline who is trying to make sense of PA’s report on the Keystone.

Woody, I suppose I should remember that not everyone works in finance and their first inclination isn’t to plug the numbers in to see if they make sense. As noted earlier, the $7.2 million is a contract number NOT a real number. Amtrak’s reported number ($26 million) is all the revenue generated by riders west of Philadelphia whether it be along the Harrisburg line or as far as NY (though it does not include revenue generated on Regionals by riders connecting from the Keystone Service). $14.12 million represents the revenue generated by the line itself plus a passenger mile based allocation of revenue through to NY, and $7.2 million represents 51% of that. I am pretty close with someone who works there and they explained it to me because I noticed PA’s and Amtrak’s numbers were so different. In other words, the amount of revenue actually generated last year varies from $7.2 million to around $27.5 million (including revenue on Regionals)depending on how you want to look at it. now Pa is estimating the route woudl increase t’s riders by another 42% on the same trains if further investments are made in the line (which would finally make it a good “high speed” corridor (or standard speed in the rest of the developed world). I tend to agree with Bruce that this bodes well for other corridors though it’s worth noting that even iff Lancaster-York-Harrisburg is smaller than the 3C’s, the 3C’s doesn’t have NY or Philadelphia on the other end of it. Together those metro areas are north of 20 million people. the drive to Pittsburgh can be upwards of 6 hours with an hour of that simply getting out of Philly and it’s suburbs…not unlike the drive to NY which frequently includes a 30 minute tunnel.

Nathanael, iirc, Amtrak is using stimulus funds to make (almost?) every station in the national system ADA compliant.

More directly, the PennDOT paper that eldondre linked to reports that “currently the stations at Elizabethtown and Lancaster are being renovated…. Public meetings have been held [about] a new Middletowne Station.” Upgrades are planned for the Exton platforms. At Ardmore, the historic station burned in the 1950s and a temporary station has served ever since. The new station is described as “shovel-ready” and planned to be multi-modal with transit-oriented residential development and new parking included.

Alon Levy on October 2, 2009 at 06:12:

“I’m not sure why you think the Ohio Hub will have an above-1 operating ratio. International experience suggests it won’t: low-speed lines like Corail are subsidized by high-speed ones like the TGV.”

Of course, the TGV itself benefited substantial from urban infrastructure that offered ordinary Express Intercity routes into the heart of big cities as well as the opportunity to run HSR services from the HSR corridor itself into the Express Intercity corridors that you are arguing should not be made available where they are most cost-effective to build.

“The closest thing to an operating profit I can find is that the Ohio Hub report projects an operating surplus for 110 mph service (i.e. not what they’re building), assuming unrealistically high ridership numbers.”

Of course 110mph is what we will be building. The service will not start at 110mph, but no reconstruction of the infrastructure we will be building is required to raise it to 110mph.

I do understand that you can frame a trip modeling comparison that makes the ridership modeling look overstated, especially if you compare ridership rather than passenger miles and act as if Amtrak NEC services were the sole rail service along the Northeast Corridor.

However, Amtrak is NOT the sole rail service along the Northeast Corridor. Fiscal 2008 ridership on the Acela was 3.4m, Northeastern Regional ridership was 8.3m, and commuter ridership on the NEC averaged about 750,000 per weekday, which puts ridership on the NEC well above 100m – but of course does not have anywhere near that impact on train miles. And the NEC is competing against public subsidized regional bus services as well.

Comparing passenger miles on the NEC to the projected passenger miles in the Triple C corridor is what one would do if interested in talking seriously about intercity transport.

And what do you present here in place of looking at transport markets, trip speeds, and available transport alternatives, in the context of modeling that has been successful elsewhere in the inland Northeast? Magical thinking: “Those lines are run to similar specs as what the FRA is calling Regional HSR or Emerging HSR, but since they’re not actually HSR, they’re too slow to attract enough riders.”

So its not how fast the trip is compared to alternatives, the size of the transport markets and the number of transport markets that the service competed in: its simply a matter of how fast the train is when it is going at its fastest. If it goes above some magical number, it will generate an operating surplus, and if it goes below that magical number, it will not.

Out here om the real world, if Express HSR can generate an operating surplus and conventional rail can not, then there is a threshold of trip speed between the two, and the threshold will be different for different corridors with different population centers at different distances apart and a different range of subsidized alternatives available to travelers.

While the belief in a magic threshold that applies universally no matter what the social geography along the corridor certainly simplifies the task of judging rail corridors, that’s at the expense of being false.

In the 79mph Cleveland to Berea, 110mph Berea to Columbus to Dayton to Cincinnati schedule, Cleveland to Columbus express is 1:38, Cincinnati to Columbus express is 1:50, and Dayton to Cincinnati and Columbus is 0:55. Three 1m+ metro areas placed under 2 hours apart, with the less populous of the pairs having a 500K+ metro area under an hour away from each, and you look at ridership modeling that predicts those services will attract riders as evidence of the flaws of the model?

It seems more likely that you are finding the ridership modeling flawed because it disagrees with your prejudices.

Just precisely as the HSR trolls do when trying to argue that the California Express HSR could not possibly be worth the investment.

“The threshold speed of profitability is probably much higher than 110 mph. On the NEC, connecting larger cities, spaced further apart,”

How does being spaced further apart make for improved performance for an Emerging HSR 110mph line?

As you noted in the discussion on the MN 110mph line study, the balance between higher speed sections and bottleneck sections means that the NEC corridor presently is equivalent to a 110mph line in operating performance now, so that before it was upgraded, it would have been cities further apart than on the Triple-C, with rail services slower than the 110mph option for the Triple-C.

“with better connecting transit and more congested roads, the line was teetering on the edge of profitability in the 1990s.”

You write as if there was no other competing transport options available to travelers in the Northeast Corridor as a result of the higher population density.

“This suggests the break-even point is about 125-135 mph”

It does indeed suggest that if you have a line with extended substantial bottleneck sections and are relying on the higher speed sections to compensate for the delays elsewhere, you need a higher maximum speed speed than if you can attain and maintain the maximum speed for more of the journey.

But the fact is that with cities spaced farther apart, in crowded right of ways with little room for new track, and facing far more common carrier competition in the intercity transport market, the passenger revenues of the Acela route was 129% of operating costs.

There is no reason why corridors with similar effective trip speeds and where due to much less road congestion they are the only transport option other than air that is time competitive with driving should be unable to capture 2% out of 3% total non-auto transport demand in a market like the Triple-C corridor.

And if it performs at that level, it will have substantially better operating ratios than presently achieved in the NEC.

I have no doubt that the conventional rail Triple-C will require subsidy, because it will be offering trips Cleveland/Columbus and Columbus/Cincinnati that are over two hours, but it will be achieving operating ratios substantially better than the Keystone corridor, and getting services up and running will accelerate the ridership growth as segments are upgraded to 110mph maximum speeds.

““How does being spaced further apart make for improved performance for an Emerging HSR 110mph line?”

It makes it more competitive with cars and buses. This is especially true in areas with gridlocked highways, such as the Hudson River crossings.”

Competing against bus services that exist because of the population densities. Once again you have a one-sided view, including factors that lower elasticity, ignoring factors that raise elasticity.

And what does it does with the competition between Emerging HSR 110mph lines like the Ohio Hub and, effectively, the Acela and air travel? What would happen to the Acela’s share of the NYC/Boston transport market if it was a genuine 150mph train service rather than a slower service that includes a few 150mph sprints?

“As of 2005-6, Amtrak had 14% of the NY-DC transport market. Getting 2% of the much smaller Cleveland-Columbus-Cincinnati market wouldn’t get you to 9 million passengers a year. The idea that Ohio alone can support 9 million with a 2-3% share is fantasy.”

Yes, its fantasy. After all, you just made it up – or rather, read one number and pretended it applied to something else because it confirmed your conclusion, decided upon in advance.

After all, the actual figure is under 2.6m, and nobody could confuse under 2.6m for 9m.

You appear to have generated your fantasy by adding Pittsburgh/Cleveland, Pittsburgh/Columbus, Cleveland/Toledo/Detroit, Cleveland/Buffalo/Toronto and Columbus/Chicago, together with incremental passengers onto the MWRRS. on top of Cleveland-Columbus-Cincinnati, and then assuming that the aggregate applied to the Triple-C corridor alone.

Aiding the distortion is the bad habit of comparing ridership numbers directly without adjusting for average trip length. The average trip in the Ohio Hub modeling is under 140 miles. Since the average Acela trip length is around 175 miles, a ridership to ridership comparison inflates the predicted transport on the Triple C by over 25%.

“At 2% market share, 9 million for the full Ohio Hub is fantasy, too. The total size of the markets in the hub is still lower than just New York-Washington. NY-DC is about 4 times 3C using a gravity model with cost proportional to travel time (this is favorable to 3C, which has shorter distances, since cost also includes access and egress times), and slightly less than twice the total for all four Ohio Hub corridors, which are together projected to get 6 million riders a year.”

The total intercity transport market that is used as the basis of the regression analysis in the Ohio Hub study is not a gravity model – its direct estimates of the total intercity transport market.

Anyone who has ever done any travel matrix regression analysis – indeed, anyone who has ever done any serious cross section analysis of any sort – knows that there is always a lot of variance in the overall regression that is not explained by the model, but is due to origin-destination pair specific factors – eg., the enrolment of Cleveland based students at Ohio State, or the state government policy on state employee travel to different parts of the state.

What you are saying is that using a privately specified back of the envelope model with absolutely no control variables, you do not get the same result as a fully (and publicly) specified regression model based on the actual trip matrix of the corridor in question, and that all discrepancies are quite obviously due to the flaws in the handling of the regression analysis in working through the COMPASS model.

It is prima facia highly unlikely that your back of the envelope modeling is sufficiently fully informed to generate the tight confidence intervals that you believe that it has.

Indeed, you seem to be doing gravity modeling based on c. 2008 populations to critique projected shares of 2025 transport markets, without noting the fact, because a hidden zero-population-growth assumption gives a result closer to the result that you want.

Since you have not defined your terms with any precision, there is no reason to be confident that you are talking about 2% of the same total.

And based on your revealed preference for “mistakes” which exaggerate the basis for the claims you make, it seems rather more likely than not that you ARE talking about 2% of a narrower market definition, because working with the wrong market definition makes it easier to confirm your initial prejudices, that the most densely populated US corridor without existing rail services must certainly be a terrible place to invest in rail infrastructure, and Rapid Rail is always a waste of money.

“Cleveland and Detroit both have negative metro population growth. Using 2008 estimates makes ridership look higher, not lower.”

This is a perfect example going directly from cherry picked data to the original prejudice without subjecting that conclusion to a moment of critical reflection.

Since Ohio has positive population growth, if Cleveland has negative population growth, that implies that there are other areas that are growing by a sufficient amount to more than offset that. And if Cleveland is the largest metro area, then in gravity model that leveling out of populations as the total expands would increase the transport market by more than the rate of population growth.

And the strongest population growth in OH between 2000 and 2008 is the Columbus MSA at the center of the Triple C, with 9.95% growth, and the second is the Cincinnati MSA at the opposite end of the triple C, with 7.24% growth.

“You should note that the TEMS study says nothing about where its numbers come from. SNCF is open about how it estimates the value of time and which models it’s using and when. TEMS just gives unsourced results, without any evidence that it knows what it’s doing.”

How can I note it when it is not true. TEMS publishes its complete zonal population, employment, and income data set, the complete functional form and methodology of its regression model, its travel survey instrument, and its broadest OD traffic matrix.

Knowing that, “noting” that they simply declare results with no information on the information sources and modeling behind the results would be a lie.

“Yes, there are intangibles, like Clevelanders studying at Ohio State. There are intangibles between New York and Washington, too – Congressmen need to spend a lot of time begging for money in Manhattan, and finance executives need to spend a lot of time lobbying for favorable regulations in Washington.”

I take it your definition of an “intangible” is “any factor, measurable or otherwise, that results in a deviation from the transport market predicted by a simple gravity model”, since there are a wide range of measurables that you omit – assuming that 22 million residents of the multi-state “Greater New York” have the same average access/egress times to Acela stations as 3 million residents of the much smaller in area “Greater Cleveland” would have to NE Cleveland, downtown Cleveland and Hopkins airport is by no means an assumption that automatically favors the Ohio Hub in your modeling.

That is why it is a substantial failing of your methodology that you have failed to do is to compare the total size of the present, existing transport markets to the sizes predicted by your model.

The COMPASS model is calibrated to the existing transport matrix, with changes from the existing transport pattern driven by the regression of correlation between measurable factors and existing transport patterns. So if the current situation is fed in, the current transport pattern will result.

By contrast, your model is an entirely uncalibrated back of the envelope estimate: “Market size is proportional to (p_a)^0.9 * (p_b)^0.9 / d^2.

This is the total size I’m taking 2% of.”

In other words, not 2% of an actual market, but 2% of a hypothetical market. And a hypothetical market where Chicago has a population of 0m.

By contrast, your model is an entirely uncalibrated back of the envelope estimate

It uses numbers calibrated by SNCF. Forgive me for trusting companies whose ridership estimates have held more than world-class consultants who’ve never been put to the test.

I take it your definition of an “intangible” is “any factor, measurable or otherwise, that results in a deviation from the transport market predicted by a simple gravity model”, since there are a wide range of measurables that you omit – assuming that 22 million residents of the multi-state “Greater New York” have the same average access/egress times to Acela stations as 3 million residents of the much smaller in area “Greater Cleveland” would have to NE Cleveland, downtown Cleveland and Hopkins airport is by no means an assumption that automatically favors the Ohio Hub in your modeling.

On the contrary: New York makes it easy to get by subway or commuter rail from where you live to Penn Station, or Newark, or Stamford. Washington makes it easy to get to Union Station by Metro or commuter rail. In Cleveland, you’d have to park at the station and pay downtown parking rates, or drive all the way out to North Central Ohio, in which case you might as well drive the whole way.

If the Ohio Hub were conceived as statewide commuter rail, I would have fewer objections to it. But it’s not; it’s intended as Ohio’s Acela, not its New Jersey Transit.

How can I note it when it is not true. TEMS publishes its complete zonal population, employment, and income data set, the complete functional form and methodology of its regression model, its travel survey instrument, and its broadest OD traffic matrix.

The travel survey form means that TEMS is projecting based on what people say they’ll do, rather than what they actually do. In its four-corridor study, SNCF stresses that its data comes from real-world observations rather than surveys, which are unreliable.

The O&D matrix, which claims that Chicago-Fort Wayne is going to be the busiest city pair in the system, doesn’t inspire a lot of confidence.

“It uses numbers calibrated by SNCF. Forgive me for trusting companies whose ridership estimates have held more than world-class consultants who’ve never been put to the test.”

SNCF is not attempting to estimate ridership for Rapid Rail, and neither do they claim that their modeling is attempting to estimate the market for Rapid Rail.

The list of assumptions you are happy to make simply because it confirms the conclusion you prefer to reach continues to grow. It is quite like reading an O’Toole attack on HSR if it were to be issued in the format of serial excerpts.

“The travel survey form means that TEMS is projecting based on what people say they’ll do, rather than what they actually do.”

Unlike your model, their baseline is what people actually do. They give you the zone matrix, if you want to run their regression with fewer explanatory variables, you are free to.

“In Cleveland, you’d have to park at the station and pay downtown parking rates, or drive all the way out to North Central Ohio, in which case you might as well drive the whole way.”

Overnight parking at Cleveland Amtrak Station is paid, daytime parking is free. And even in the conventional rail starter system, the “next station” is the RTS West 150th and Puritas Avenue station. In the first, “Cleveland Hub” layer, there is also a station in Summit County at Cuyahoga Falls, and one near Mentor.

Another false assumption made because it simplified confirming your original prejudice.

For: “On the contrary: New York makes it easy to get by subway or commuter rail from where you live to Penn Station, or Newark, or Stamford. Washington makes it easy to get to Union Station by Metro or commuter rail.” … first, this is still a greater time of travel for the average resident of the 22m Greater New York metropolitan area or the average resident of the 8m Greater DC metropolitan area to the closest Acela station.

If you can just arbitrarily allocate the entire metro area population to “convenient access to the closest Acela station” by virtue of the access of a smaller population that have that access, then I could use Cleveland neighborhoods with easy access to the Red Line at Puritas and Blue and Green lines downtown to allocate all of metro Cleveland to “easy access”.

Of course, I wouldn’t do that, but you just did. So, again, another false assumption made because it simplified confirming your original prejudice.

On a completely different note, apparently the state of Ohio estimates that,

the best possible train run time from Cleveland to Cincinnati would be 5:20. This does not include any time for station stops, slow orders, recovery time, and time needed to hand-off the passenger train from one railroad to another. For the purposes of continued planning and coordination with the railroads, a total train run time of 6 hours and 30 minutes has been assumed in the schedule development.

The TEMS study, which estimates that the trip would take 3:50, is in other words making unrealistically optimistic assumptions about speed. Not very encouraging.

I once rode this train from Richmond VA to Lancaster PA and it was a good ride. The hard part of it was when I spent three hours at the Philli station waiting for a train to go to Lancaster. Once the train that was going to go to Lancaster came it had to spend a hour changing from eletric to Oil power to go to Pitsburg which ate up one hour. If they extened the catenary wires west of Harrsionburg it Pitsburg and add a few more trains running on it it would have knocked down three hour wait down to a half a hour. When I did get on the train to Lancaster it was packed to the gills which in away is good for the train it showed there were a lot of riders for it. But if they added more trains more people could fit on it to in that the conductors were turning people away after we got on the train. So more trains would carry more money.

Another thing I wounder about when they extend the catenary into Pitsburg would they be able to use the same Classic catenary masts but with a up dated retro look to them such as they can have the same top set of wires on them to get eletric power from wind farms and solar farms. The classic catenary masts currently get their train power from renewble hyrdo power off the Susahanna River from a 1920’s power station. They could expand this hydro staion by devoting more generators to catenary power production for the new high speed rail lines. The part of the classic 1930’s masts and the new classic based masts could have the modern wire hooks ups on them that would allow them to raise train speeds but keep the look of the classic masts.

I do wonder why Amtrak continues running the Pennsylvanian between New York and Pittsburgh. If it ran between Harrisburg and Pittsburgh the same equipment and the same crew resources (not necessarily the actual crews) would suffice for three roundtrips a day instead of one. Since between New York and Harrisburg the Pennsylvanian is slower than the corridor trains and has to change engines at Philadelphia to boot, surely no one takes it to travel between New York and Harrisburg. Three chances a day for a two-seat trip is surely better than one chance a day for a one-seat trip.

With some judicious equipment swapping with the Capitol Limited when it passes through Pittsburgh, Amtrak could arrange for scheduled maintenance to be performed at Ivy City without expensive dead-heading.

What would be the downside?

Alon has the sharpest answer.

But there could be other factors. Maybe NS didn’t want two more daily passenger frequencies on its tracks Harrisburg-Pittsburgh. Before the Great Recession knocked traffic off the rails and roads, that was a pretty busy line. Maybe Amtrak needs to keep the one diesel train on the Keystone route because of a shortage of electric equipment.

Actually, as Alon knows, they wouldn’t REALLY have to have timed transfers. Anyone who gets on board for an Amtrak trip of any kind should certainly expect delays.

I think it’s a great suggestion.

The Pennsylvanian leaves now leaves Philly at 12:42 p.m. and Harrisburg at 2:26. But by 9 a.m., already three Keystone trains from Philly have reached Harrisburg. One mid-morning departure could get into Pittsburgh well ahead of the scheduled 8 p.m. arrival of the Pennsylvanian, in time to meet business colleagues, family, or friends for dinner. A departure heading west around 5 or 6 p.m. would allow a full day of work in Harrisburg, arriving in Pittsburgh only a little past the usual bedtime.

The scheduling of the Pennsylvanian needs a fresh look, no doubt.

The Pennsylvanian is among the most reliable of Amtrak’s trains off corridor. I would think a twice daily Philly-Pitt would be more desirable, many connections to DC and NY can be had in Philly. Perhaps the biggest problem with this is that its all in one state so Amtrak needs state funding. Perhaps another alternative is a twice daily NY-Pittsburgh train that runs via the Pittsburgh subway skipping 30th st and stopping only in North Philly. the loco change could occur in Harrisburg. that should eliminate maybe 30 minutes from the schedule in Philly (actually more since they have such long dwell times in Philly) plus another ten or so minutes on the way to Harrisburg since it’s electric. The 15 minute layover in Harrisburg would then be used for the locomotive change. If the train runs with a cab car, it should allow Amtrak to reduce the time needed to turn the train since it would no longer need to wye. I think Amtrak could negotiate another 30 minutes in trip time reduction with NS as the train often arrives early. anything more would probably necessitate investment in tracks, signal, and interlockings.

It used to be four tracks, yes. It has been cut down to three since, I think in the Conrail days.

Amtrak has release a PRIIA mandated study on adding service to the Pennsylvanian. http://www.amtrak.com/pdf/PRIAA/PennsylvaniaServiceStudies.pdf

@75, the report says there are two mainline tracks over most of the route between Harrisburg and Pittsburg.

eldondre — Answering my own question from that study on Pennsylvanian services:

“Additional Harrisburg- Altoona-Pittsburgh service would have to be state-supported. Section 209 of the Passenger Rail Investment and Improvement Act of 2008 requires Amtrak and states to develop and implement a uniform methodology to allocate operating and capital costs of existing and future Amtrak routes less than 750 miles in length, and that, by 2013, all states pay an equivalent share of the costs of such routes that are not covered by farebox revenues.”

At the current state of the tracks, it’s probably easier to serve Cleveland and points west on the Water Level Route. It’s 60 miles longer than the PRR route, and is straighter and flatter, giving it more incremental upgrade potential.

On another note: I find it really annoying that Amtrak’s supposed to live on only state subsidies for its short-distance routes, but not for its long-distance ones. It turns Amtrak from a railroad into pork service for Montana and New Mexico.

I think the study was offbase. they should have looked first at what would be required to run a twice daily with the same two sets of equipment. Track restoration would be good for both passenger and freight operation. One would hope that portion of the study gets pushed forward under an intercity rail grant but first they need to look not just at adding tracks but speeding up the trip (which PA submitted for funding of an HSR study). It is frustrating that this is oen of the few projects that is recomended yet it’s ineligible for federally funded services.

Alon — You’ve pinpointed another problem that Team LaHood hasn’t revealed the answer to.

I guess Cali’s HSR will be tax-exempt. But in Illinois they are talking about using existing ROW for the first semi-fast 110 mph route Chicago-St Louis. So those trains will run on taxable tracks, unlike the planes landing at O’Hare. Then if or when they go with the easterly route for 220 mph service, and put in new dedicated tracks in existing ROW, who will pay taxes on that?

In fact almost all the “incremental” upgrades where freight will run on the tracks as well invite further taxation on the federally funded improvements.

I’m sure the schools could use the funds, but I fear the new tax money will go to improving the local streets and roads.

I don’t have a problem with railroads paying property on their right-of-way, but it is absolutely senseless to disincentivize investments in improving the infrastructure. If you have a class 1 mainline, it ought to have one valuation irrespective of how many tracks, signals, turnouts, etc. are on it. Same thing for the regional RRs and short lines. There’s nothing wrong as I see it in getting more property tax for the locals from yards, transload facilites, office buildings, depots and other uses when they’re improved.

Of course another way that the railroads wring cash out of their free land is by selling off utility easments, that then make re-establishment of tracks problematic.

Ocean R — To get to 220 mph HSR through PA, we’ll be looking at some magnificent tunnels, with costs to match. And above, Alon points out that incremental improvement Chicago-Cleveland-NYC will probably cheaper on the Water Level Route. One day we’ll want HSR on Philly-Pittsburgh, but that day probably is a long, long time away.

Meanwhile PA has some plans to chop a few hours off the conventional service. That’s fine — but I’m with you. Let’s leave the Horseshoe Curve in the schedule. It could slow us down 5 or even 10 minutes, a very quick detour to see a scenic, historic, and engineering wonder not to be missed.

Actually, Pitt is already shorter than the water level route and with tunnels, it would be even shorter. I don’t think a new high speed segment would necessitate abandoning the old route, you’d just end up with two different levels of service (acela vs regionals for example). you tart by upgrading the current ROW, then work on. When talking about reak HSR (be it 220 or 186) I think th ePitt Phily route makes more sense in that the pop centers are more centered. Philly is bigger than all of western NY stops combined and pitt is twice as large as buffalo, all by making only two stops. I’d also argue that Pitt-Cleveland is a larger market than Buff-cleveland. nonetheless, I’d be happy, for now, with a reasonably competitive conventional route