30 October 2011

Blended Like Oil and Water

The Caltrain-HSR blended proposal, initiated by State Senator Simitian, Congresswoman Eshoo and Assemblymember Gordon to break an impending political impasse on the peninsula rail corridor, has now been evaluated by the HSR Authority's consultants.

Their version of it, submitted in a recent report to the legislature, would cost an astonishing $5.3 billion, even more than the $4.2 billion that the entire peninsula section was estimated to cost back in 2008.  This stunning price tag could have several possible explanations:
  1. a repudiation of the blended approach, accomplished by deliberately inflating the budget.
  2. an effort to pour the largest possible amount of concrete, regardless of actual operational benefit.
  3. sheer engineering incompetence and complete disregard for the new fiscal reality.
A Closer Look At Costs

While the detailed breakdown of that project budget is not provided, it's not too hard to back it out based on what is described.  Here's how it might approximately add up, including project management and contingency costs:
  • $0.3 billion to build grade separations at 25th, 28th and 31st Avenues in San Mateo, and to expand the corridor to four tracks from the southern border of San Mateo up to the vicinity of 9th Ave (milepost 18.3), with new four-track stations built at Hillsdale and Hayward Park
  • $0.3 billion to expand the existing grade separations in Belmont and San Carlos to four tracks.  This would include new stations built at San Carlos (south of the current location) and Belmont.
  • $1.0 billion to fully grade-separate the rail corridor through Redwood City, with expansion to four tracks and an elevated station.  New grade separations would be created at six locations: Whipple, Brewster, Broadway, Maple, Main, and Chestnut.  The new four-track section would tie in to the existing four-track section at Redwood Junction.  Should the Redwood City grade separations be built below grade, costs would be even higher.
  • $0.5 billion to burrow a single-track tunnel under the Millbrae station to "save" the existing Caltrain / BART station from complete demolition.  As described in the Alternatives Analysis, the station itself would be re-arranged to accommodate a segregated HSR platform at grade and the southbound Caltrain platform underground.  The tunnel approach would require two new grade separations at Center St. and Santa Paula.
  • $1.0 billion to build a six-mile (yes, six miles!) 60-foot tall elevated viaduct from Lawrence Expressway (milepost 40.9) all the way into the upper level of a massive new elevated HSR station complex in San Jose.
  • $0.2 billion to demolish existing overpass grade separations at De La Cruz (Santa Clara) and Hedding (San Jose), to be replaced with underpasses to make way for the six-mile viaduct.
  • $1.5 billion to electrify the entire corridor, an estimate based on Caltrain's latest electrification budget but discounting the cost of Caltrain's new electric train fleet.
  • $0.3 billion for positive train control and technical integration with the HSR system's train control system, which will differ from the PTC solution adopted by Caltrain.
  • $0.2 billion for reconfigured station facilities at San Francisco and San Jose.
TOTAL:  $5.3 billion

All of these individual investments must be examined in the context of their operational utility, i.e. the value they add to creating a blended Caltrain / HSR rail corridor that is as flexible and efficient as possible and best meets the service expectation of rail passengers.  And on that basis, most of the above list falls woefully short.

Let's Do Some Value Engineering
  • The six-mile (yes, six miles!) 60-foot tall viaduct to approach San Jose adds little operational value compared to cheaper alternatives such as laying additional track at grade from CP Coast (milepost 44.6) into San Jose.  The corridor is mostly 100 feet wide in this area, so the need for a viaduct--let alone a six-mile long viaduct that requires demolishing some perfectly fine grade separations that already exist--is highly questionable.  It is almost an insult to Simitian et al., who specifically requested that aerial structures be minimized.  This viaduct is the outcome of lazy engineering, the sort that avoids inter-agency coordination issues by using megatons of concrete to build over Caltrain, ACE, Amtrak, VTA, BART, UPRR, Caltrans, and everybody else.  Bottom line: with some hard negotiations and minimal takes of a few slivers of industrial property, four tracks can be built at grade all the way into Diridon Station.  Savings: - $1.1 billion
  • A four-track mid-peninsula overtake facility is the key enabler of a blended solution. However, the bulk of the cost of this mid-line overtake is a massive Redwood City grade separation project that would eliminate one of the clusters of grade crossings on the peninsula rail corridor.  Is it truly necessary to do so right away in the first phase of the blended project?  Consider these two options:
    1. The CHSRA's proposed overtake facility: 9 miles long with 5 stations (potentially including the HSR stop at Redwood City, which does not help with overtaking)
    2. A somewhat shorter overtake facility: 6.5 miles long ending at Whipple Ave (milepost 24.8) with 4 stations.
    Option 2, while only 3/4 as long, entirely avoids the need to grade-separate an expanded four-track corridor clear through downtown Redwood City.  It also delays a fight against the city's inevitable demands to burrow the rail corridor inside an astronomically expensive tunnel.  The shorter overtake might make operations slightly less flexible and robust, but at the very worst, every Caltrain could be held or slowed for just two minutes to make up for that.  Bottom line: with minor timetable adjustments, the lion's share of the cost of the mid-line overtake can be avoided.  Savings: - $1.0 billion
  • The half-billion tunnel in Millbrae is the wrong answer to the question of how to fit four Caltrain / HSR tracks through this station.  The entire Millbrae complex, including 3000-car parking garage, cost about $100 million in today's dollars.  Portions of it can be torn down and reconfigured with four tracks at grade for far less than that sum.  Bottom line: whatever the engineering constraints, you simply don't build a $500 million tunnel to "save" a $100 million station--at worst, you tear it down and start over.  Savings: - $0.4 billion
  • The need to integrate Caltrain's CBOSS train control system with the HSR train control system will drive unnecessary costs, most likely resulting in what is known as "dual fitment" of train-borne signaling equipment.  Each high-speed train, California-wide, would need to be fitted with CBOSS hardware and software, with the appropriate interfaces to allow a seamless change-over when entering or leaving exclusive HSR tracks.  This is not a trivial expense, since safety-critical signaling computers approach the cost of aircraft avionics.  Bottom line: deploy ERTMS, not CBOSS.  Savings: - $0.2 billion
TOTAL SAVINGS: $2.7 billion

The very bottom line is this: we can get 95% of the bang for 50% of the buck.  Somebody needs to inject a little bit of sanity into the planning process if a blended solution is ever going to be feasible, if for no other reason that in these times, affordability determines feasibility.  That's why the CHSRA's proposal for the blended system is a disgrace, larded as it is with operationally dubious viaducts, tunnels, bridges and underpasses; in short, a project dreamed up by civil engineers writing their own checks.


  1. Of the remaining $2.6 billion, $1.5 comes from electrification. Per kilometer, it's about the same as building a full-fat LGV, including the electrification, four times as high as old plans to electrify Metrolink, and about seventeen times as high as the cost of the Sables d'Olonne electrification project.

  2. "Only" a 1.0 billion pittance for six mile of urban elevated -- elevated above an active rail corridor, elevated above high traffic freeways, elevated above existing residential and commercial uses -- PLUS the most iconic pan-galactic multi-level station in the universe? All in a city which wants and expects and always gets "no expense spared"?

    TRIPLE that number to start, and go up, WAY up, from there.

  3. $100 million per km of el sounds about right to me, actually. The Tohoku Jukan el-over-el (in the Tokyo CBD, to boot) is $150 million/km, by comparison. And the Chuo Line grade separation project was $1.8 billion for 13 km, of which 4 are in a trench and the rest elevated. Japan tends to be a high construction cost country, though not as high as the US.

  4. I may have slightly miscalibrated the relative costs of these two items. Say 1 bil for electrification and 1.5 bil for the viaduct, and presto I can save you another half-bil!

    This kind of stuff makes it really hard to be a supporter of HSR.

  5. Alon,

    Based on what we see around here, today, they could barely even build the SJ Station Iconic Canopy for a billion. Compare Transbay.

    They'll easily blow a billion on Didiron Hypergalactic without any connecting tracks.

    And their six miles of elevated structure for a billion? In San José, Capital of Silicon Valley, Capital of fraudulent PBQD-promoted rent-seeking rail "transportation" scams? In California? Compare to a third of a billion or so for the trivial (anywhere else in the world) couple miles of grade separation in San Mateo Not going to happen!

    There's no way the HSR-only fuck-you-Caltrain Tamien-Santa Clara stilt-a-rail is going to be delivered for less than a couple billion US dollars. No way. Just look at the parties involved! Every single one of them, without exception seeks cost maximization at every point.

  6. For crying out loud! This report budgts _more_ to integrate HSR advanced ATP with CBOSS.

    For pete's sake, why not just kill CBOSS? What does it buy over existing, HSR-proven, advanced ATP systems? The ability to open at-grade crossing gates at stations in front of stopped trains??

  7. @clem: you state $1.1 billion savings from eliminating the aerial concrete overpass into the "Diridon Intergalactic' station in San Jose.

    How did you reach that figure? Is it ($1.0 bn for viadict + $0.2bn for demolishing existing overpass) - ($0.1 bn for eminent-domain takings), or what?

  8. I figured it as:

    -1.0 viaduct
    -0.2 De la Cruz & Hedding grade seps
    +0.1 quad-tracking within existing corridor, and minor industrial land acquisition.

    Units are BILLIONS, so that last little bit is still a hundred million. Not chump change.

  9. @clem: thanks, it was that last $0.1bn I was asking about, and I agree $100m is a nontrivial sum.

    But what about the electrification costs? $1.5bn? the Auckland, NZ electrification contract was NZ$ 80m. I can't find the total catenary length on-hand, but the initial (2007?) study was for 170km. Wikipedia says the Auckland project will use 3,500 masts. At 50m (metre/meter) separation (LGV standard is a tad more than that), that comes close to 170km.

    Even adding in 2x for substations, and maybe 2x for the speed difference, there's almost a factor of 4 lower than the $1.5bn numbe4. What's up with that??

  10. For comparison, is there a cost estimate for BART between Millbrae and Santa Clara? That would eliminate the "blending" problem. The Transbay Terminal would be more manageable. In the case of bottlenecks, tunnels for BART trains would be cheaper than the proposed HSR tunnels, which seem sized for freight trains. Of course, the rails are spaced too far apart for Roman chariots, so BART and HSR would be completely separate systems.

  11. @peter; oh, but if only Caesar hadn't "decided to proceed with his advance into Britain, because he realized that in all of the Gallic wars, help had come to our enemies from there"[1] ... Britain's roads might have had different rut widthds, leading Stephenson to pick a different gauge.
    But, surely not indian broad-gauge!

    BART estimates are not useful. Budgeting for PART extensions includes the cost of additional trainsets, since BART trainsets are used to their limits. So you'd have to find, and deduct, the cost of trainsets. And dollars to donuts, any BART extensions will follow the existint BART "House Style": stations in Brutalist concrete, segregated after fare-getes, etc., etc.

    A sane "blended" approach would have tracks at Cahill St station at the current grade level. ;)

    [1] De Bello Gallico, Julius Gaius Caesar, bk IV.

  12. Agree, 1.5 bil for electrification is a ridiculously inflated sum. Like everything else. The fundamental issue is that we are seeing the emergence of a transportation industrial complex, very much like the military industrial complex and using the same tried and true playbook to transfer enormous sums of public wealth into private pockets.

    For example: massively inflating the cost of a blended solution, which in principle ought to be far cheaper, makes a BART solution increasingly attractive.

  13. What I read about the Auckland electrification was that it would cover 80 km of route. I presume we're both working from this story; Phase 1 is 80 km in total.

    On another note, can we let go of the myth that standard gauge comes from Roman chariots? In 1820s' Britain, there were already many industrial railways, operating on a variety of gauges, and it was well-understood that broader gauge offered more stability, and narrower gauge offered lower construction costs and easier curve turning ability. The Manchester and Liverpool and connecting lines (i.e. most) were built to 4' 8.5", but other main lines had different gauges. The Great Western used 7' 0.25" gauge for decades before giving up and converting to what the rest of Britain ended up with. The South Carolina Railroad and the other Southern railroads used 5', but ended up outcompeted by Northern industry and had to convert.

  14. This comment has been removed by a blog administrator.

  15. Richard's comments, sanitized... Please don't make me do this.


    Auckland electrification is 80 route-km, 175 track-km. Masts have gone up all over the place and wiring is about to begin. The EMU contract was awarded (there were 11 bidders) a couple months ago. (This unlike Caltrain, where, should the (deleted) running the operation ever manage to issue a bid package, the fix will be in way in advance and most won't waste their time.)

    Auckland is also installing ETCS-1. This unlike Caltrain, all of whose staff and consultants richly deserve to (deleted). (Note that most of the big experienced global signal companies didn't bother bidding the the Caltrain CBOSS technical abortion and massive fiscal fraud, because the fix was well and truly in.)

  16. Well, 3,500 masts, at 50 meters between masts, gets you (3,500 * 50) = 175,000 metres or 175km of catenary. With juntions and more masts on curves, that's in pretty close agreement to 80 route-km.
    Those are the numbers I worked with.

    I got my data from the Auckland Regional Authority web pages.
    I also followed the Kapiti Coast electrification -- I used to live there.

    Even quadrupling the cost, that's more than 4x cheaper than Caltrain's estimate; almost 5x.

  17. Adirondacker1280001 November, 2011 23:45

    Richard's comments, sanitized... Please don't make me do this.

    It lives on in RSS feeds.

    Interesting little tidbit over on Railroad.net today. ETMS and ACSES are inter-operable. Apparently the Class 1s have all agreed ETMS will be the standard. Nice diagrams showing that the radio portion of the whole thing can be almost anything. Very ERTMS like. I wouldn't be surprised if the other bits are swappable with ERTMS equipment. Like running OSX instead of Windows on you x86 computer. Probably more like running Xbuntu instead of Puppy.


  18. Which makes the CBOSS lunacy all the greater...

    What really frustrates me with this whole stinking pile of is that the ideal solution would be either (a) to call in some real rail experts from the UK, France, Germany, Sweden, Japan, Korea, and even Australia to put a serious value-engineering scalpel to this whole thing or just dissolve CAHSR as an agency and have a new one begin with the current Business Plan with the two edicts Cut Fat and Make It Work.

    But we can't do the first 'cuz WE'RE AMERICA, BABY! Why should we listen to what experts around the world who've made the thing work have got to say? We can reinvent the wheel just fine!

    And we can't do the second because with the lunatics in charge at the only level which supersedes the state they'd just use it as an excuse to abandon HSR altogether.

    'Tis too bad by far the sanest project was the first axed. Fortunately, the governor up there seems to have made it his political mission to not just destroy his own political career on a state level, but so totally tarnish his party's reputation there they'll cease to be relevant for a generation or more.

    ...Yes I'm feisty. Geez, I'm starting to sound like Richard now, aren't I?

  19. That was a pretty enlightening presentation on ETMS. The coolest safety-critical feature is that it will automatically sound the horn at grade crossings, which will make about half of the job of the locomotive engineer, as defined by the FRA, redundant. Now it just needs cameras to automatically spot a red flag on the tracks, which will allow it to pass efficiency tests, and render locomotive engineers completely redundant.

  20. @Steve—I just replied to someone on The Transport Politic and thought the same thing. In the space of a week I’ve gone from thinking CAHSR was a worthy (if flawed, but with room for improvement) project to a long con.

  21. Though this report is definitely a setback and made me a lot more wary of the project, I haven't completely jumped ship, if only because the CV segment has a relatively reasonable design. The problems appear to be in Palmdale and the entire Bay Area-Central Valley segment; the former could be avoided with Tejon, and the latter will likely be delayed to the last phase anyway, when perhaps saner people will be in charge.

    On a side note, I think that Altamont's chances have actually slightly incresed from this debacle. Somebody's gotta notice something is horribly wrong in SJ, and that there exists an alternative that takes out the useless viaduct...

  22. I've got a few sincere quesions:
    1: Is the idea of HSR up the Peninsula to have as many stops as Caltrain?
    2: If all stops are supposed to be the same, couldn't you just do away with Caltrain all together since that's about the current frequency of the trains?
    3. As far as all this planning goes, is the HSR Authority comprised of Americans, who have little to no experience with this sort of project?

  23. "1: Is the idea of HSR up the Peninsula to have as many stops as Caltrain?"

    No. Moreover it should be the plan that there are a (very small number) of different types of Caltrain service that have fewer stops than Caltrain locals.

    This mixing of any-stops and fewer-stops services is done everywhere in the world on every mainline (non-subway) rail network. There are ridership, operating cost and especially capital cost trade-offs involved as trains with disparate average speeds attempt to share the same tracks.

    These articles might give you some information, but there really isn't tutorial-level information here.

    "2: If all stops are supposed to be the same, couldn't you just do away with Caltrain all together "

    Yes, in some abstract sense. But you'd have trains coming in from LA in 2 hours and then spending an hour crawling the last few dozen miles. You'd also make every passenger between, say, Mountain View and San Mateo exposed to any disruption to train service at any point anywhere all the way to LA. And lastly there are going to always be two or three times as many local SF...SJ riders as there will be riders in inter-regional high speed trains. So none of this makes any sense.

    The only thing that would make sense would be to have inter-regional HS trains "slot" into the normal intra-regional Caltrain service pattern and, if there are empty seats available on the HS trains, make them available to Caltrain passengers in addition to the regular old Caltrains. (This is how things work in Switzerland, where efficiency and customer service are paramount, but the exact opposite of how things work with say Amtrak in the USA, where nobody cooperates on anything and train passengers are treated as badly as by airlines.)

    "3. As far as all this planning goes, is the HSR Authority comprised of Americans, who have little to no experience with this sort of project?"

    Yes. The CHSRA CEO is an Anglophone South Effrican, but that's about it. There's far less than zero expertise in the organization, and a massive resistance to anything "foreign" -- starting out with spending millions of your tax dollars to avoid millimeters and kilograms.

    One example: the character who was or is the "Discipline Lead" for "Operations Planning" and the "Discipline Co-Lead Engineering Criteria" was completely unaware that there are commuter trains in Germany, and knew nothing whatsoever of the world outside the USA and London. Nothing.

    The ignorance and unprofessionalism of all of the involved parties is simoply breaktaking.

  24. Have any of the regular posters here made any formal proposals to the CAHSRA about how things ought to be? What was the response?

  25. This is heartbreaking!
    I'm surprised the old hands here have any patience left at all.
    @thatbruce .... good question. Is it time to take some organized action? (like monopolizing the public comment portion of Caltrain Board meetings with thoughtful comment and critical thinking?)
    Count me in!

  26. Clem, I agree with you wholeheartly on the opportunities for value engineering, especially the Millbrae situation.

    For the segment between CP Coast and Diridon, UP/Amtrak can easily be segerated from Caltrain/HSR by having their own dedicate pairs of track all the way to Diridon, all while Caltrain/HSR use the west 2/3 of the station.

  27. heck, engineers designed a 100% or even a 120% solution that's too expensive, now it is up to the leaders to make compromises and decisions to reduce cost.

  28. "starting out with spending millions of your tax dollars to avoid millimeters and kilograms."

    I suspect this may actually be a legal requirement, to use "American" units for various things.

    Blame Ronald Reagan. We were well on our way to metricization under Jimmy Carter, and then that JACKASS cancelled it and kept the US government using archaic measurements.


  29. Adirondacker1280007 November, 2011 20:07

    I suspect this may actually be a legal requirement, to use "American" units for various things.

    Bought any 67.6280454 ounce bottles of soda lately? Or 25.360517 ounce bottle of wine? Wanna buy something nice for the holidays? How about nice 0.198129039 gallon of whiskey? A stocking stuffer for the gearhead in your life? A few sockets in the most used sizes on current American cars ( anything after 1980 or so ) in 0.354330709 inch, 0.433070866 inch and 0.511811024 inch?
    Your doctor tells you to take 10 grains of aspirin ( two 5 grain tablets ) every four hours until you feel better or three days...

  30. Clem, please submit these Value Engineering ideas for SF-SJ and SJ-CV EIR scoping.

    These ideas not just cost saving, but avoid massive soil disturbance that could release tons of toxic substance due to the corridor being an active railroad corridor for so long.

  31. The metric thing is really tangential, but consider this: every single thing, every CAD program, every bolt, every sub-assembly, every assembly, every piece of software, anything that deals with anything from the rail fixation up, will either have been designed and manufactured in standard metric units, or else will have been gratuitously re-developed in USA-USA-USA at extra cost and to the exclusion of otherwise competitive products and services.

    There's no reason at all that the Caltrain line couldn't have kilometer posts and that it should use a directly off-the-shelf signalling system that displays information in the English language and kmh, even while adjacent highways have mph speed limits. None whatsoever. Train drivers just need to make sure that one number displayed on a screen is smaller than another; they're not in the business of telling the difference between 71mph and 72mph by eyeball.

    There's no reason that Caltrain couldn't use standard CAD systems for laying out medium speed overhead power (on standard 80m pole spacings for sub-200kmh track) ... but instead they've reinvented a square wheel and using 50 YARD pole spacing that they pulled out of their own rear ends.

    And so on.

    It's all make-work. And make-work that costs over and over: first in direct make-work cost; then by reducing procurement competitiveness; by introducing errors; and by locking in to a non-competitive one-off unique home-brewed kludge-fest of "standards". Hmmmm.... sounds like the CBOSS story, doesn't it. Well, expect the same thing to be repeated over and over and over again as Caltrain and PBQD's World Class Railroading Professionals Who Know More Than The Rest of the Galaxy Combined come up with more and more and more of their own special "standards", standards in God-fearing Church-going waterboarding-condoning pounds and furlongs.

  32. On the topic of metric vs. American units: Australia converted all their railroads to metric at some point in the 1970s, including everything on both the engineering and operational sides. The distances that used to be measured in miles and chains were re-measured in kilometers, and speed limits were all changed from mph to km/h. And even British railway engineers is perfectly capable of working in metric, as evidenced by the KCR East Rail in Hong Kong, which was largely a British design and all metric from the start of its current incarnation.

  33. All UK rail engineering is pure metric. The only thing that isn't are the speeds and limits displayed to trains drivers, for reasons of massive transition complexity and cost.

    That transition complexity is simply not an issue of any kind on the minute little isolated trivial Caltrain shuttle line, nor would it be a factor on the stand-alone from-scratch HSR line, even when it shares track with post-FRA Caltrain and a post-FRA Metorlink line Sylmar-LAUS-Anaheim-Orange.

    PS The Australian "national" (!) rail conversion took longer. Highway signs were done pretty much over a weekend. (OK. I exaggerate a little.) It wasn't any big deal.

  34. In sum, the HSR nightmare not only continues but gets worse by the day. Incredible. And the state is broke with a credit rating worthy of Greece.

  35. "And the state is broke with a credit rating worthy of Greece."

    Ummm, bullshit?