26 September 2017

Thoughts on Palo Alto

There is a vigorous discussion of grade separations now underway in Palo Alto. It misses several important points:

1) Grade Separation is not one project. Trying to come up with a single, grand unifying grade separation scheme for the entire rail corridor through Palo Alto is to over-constrain the problem and to limit the range of feasible solutions. The wide geographical spacing of the four remaining grade crossings in Palo Alto leads naturally and logically to three separate and independent projects: Alma, Churchill, and Meadow/Charleston. These three projects can be and should be completely decoupled from an engineering perspective, if not from a political perspective. The underlying geometry of Palo Alto does not lend itself to a single project.

2) Creating new cross-corridor access is not grade separation. While it is understandable that the city desires to knit together neighborhoods on opposite sides of the track by creating new places ("trench caps") where people can access the other side of the corridor, this is not grade separation and should not be funded by scarce grade separation or transportation dollars. It can't be said that the city was actively divided by the rail corridor, since the rail corridor was in place decades before Palo Alto grew into a city. While everyone agrees that new cross-corridor access would improve Palo Alto, the distinction of scope between grade separation of existing crossings (today's network topology) and new cross-corridor access (tomorrow's network topology, a nice-to-have) should remain crystal clear. Muddling the project scope will muddle the discussion of funding.

3) Split-grade solutions should be studied with due diligence. When the city commissioned a grade separation study from engineering firm Mott Macdonald, the council deliberately excluded from consideration any designs where rails or roads might rise above existing grade. From the outset, this eliminated the standard solution that every other peninsula city has adopted: San Bruno, Burlingame, San Mateo, Belmont, San Carlos, Menlo Park and Sunnyvale either already have or are planning split grade separations, where the rails are raised a bit and the streets are lowered a bit. Turning a blind eye to split grade solutions, however controversial they may be, casts doubt on the entire decision making process. Without due diligence in studying a full range of grade separation solutions, the politics of assembling the necessary funding will become unnecessarily complicated.

4) Funding matters. The most expensive options are the most popular because the cost isn't yet borne by anyone. Everything is paid for with OPM or Other People's Money. If you went to a restaurant with OPM, of course you would select the Filet Mignon (or truffles, if you're vegetarian). A selection process that ignores funding is detached from reality. This also means teaching people about orders of magnitude: capturing ill-defined revenue from new uses of 45 acres of highly impaired land that the city doesn't own, even at Palo Alto prices, doesn't begin to pay for the astronomical expense of burying the tracks. Until funding is seriously factored into decision making, it's all just unicorns and rainbows.

5) County grade separation funding is always at risk. While 2016 Measure B set aside $700 million for grade separation projects, a 3/4 majority vote of the VTA board is all that it takes to re-program some or all of that funding "as circumstances warrant" towards BART, in the exceedingly likely event that the San Jose extension goes over budget. Spend it soon, or flush it into a giant sink hole in San Jose.

Failing to properly acknowledge these realities will likely leave Palo Alto's decision making process tied in knots as other cities move forward.

12 August 2017

Freeway Lanes of Caltrain

If everyone drove instead of taking Caltrain, how many more lanes would peninsula freeways need to absorb the additional traffic?

The way to answer this question is to count how many train passengers ride past any given location, in each direction, within the span of one hour. Caltrain publishes all the information you need to do this calculation rigorously, without making any assumptions: the timetable tells you when each train passes each location, and the 2016 weekday passenger count by train tells you how many people are on board that train at that time.

Four cases are considered: morning northbound, evening northbound, morning southbound, and evening southbound. Rather than picking a fixed morning and evening hour over which to count passengers, we slide a one-hour window across the peak period until we find the peak hour at each location, during which the most passengers ride past. Caltrain operates five trains per hour per direction repeating on an hourly cadence, so we never count more than five trains in the totals.

It is an easy but tedious calculation, perfectly suited for a computer.  This is what pops out:


This graph reveals many of the features noted in ridership reports: the flow is asymmetrical with more riders traveling northbound AM / southbound PM, the Gilroy branch is dead, Stanford generates enormous ridership, etc.

Translation to Freeway Lanes

To convert the number of Caltrain passengers into freeway lanes, very few assumptions are needed, and those we need can be backed up by references.
  1. A congested freeway lane operating at 45 mph can carry 2000 passenger cars per hour, according to the Federal Highway Administration's HPMS Field Manual (Parameter values: FFS = 45 mph, BaseCap = 2150 pcphpl, PHF = 0.95, fHV = 0.98, fp = 1.0).
     
  2. The average vehicle occupancy (AVO) is 1.3 people, based on two studies of the 101 corridor in San Mateo County. This figure includes buses, van pools and corporate shuttles.
This means a single freeway lane can theoretically carry 2600 people in one hour. Note this is a very optimistic figure because slight perturbations in the flow of traffic can cause slow-downs that reduce throughput due to lower free flow speed (FFS). But we'll use this very high number to make an extremely conservative estimate of how many lanes of freeway can carry all of Caltrain's ridership.

Freeway lanes typically do not change directions to accommodate peak flows. That means we must consider northbound lanes separately from southbound lanes, with no possibility of re-allocating the lane capacity to accommodate the AM/PM flow asymmetry that is observed on Caltrain. In practice, this means we must add the northbound peak flow (AM or PM, whichever is highest) to the southbound peak flow (again the highest of AM or PM) to size the number of equivalent freeway lanes. Looking at the graph above, which shows the highest flow is northbound AM and southbound PM, we must add AM northbound and PM southbound people per hour, and divide by 2600 people per hour per freeway lane. Here is the result:


So as of 2016, plain old diesel Caltrain equals about 2.5 lanes of freeway, including both directions. If you integrate the area under this curve, you get how many lane-miles of freeway would be needed to replace Caltrain. That number is 119 lane-miles. These are very conservative lower bounds.

When you hear the argument that "millions" of people use highway 101 but only about 30,000 people use Caltrain, shut it down with facts: today Caltrain amounts to 2.5 / 8 or at least 30% of the lane capacity of highway 101 during rush hour. The reply might be that not all those people would end up on 101, but with an average trip length of 23 miles, which driver wouldn't use a freeway?

Future Capacity Implications

Caltrain capacity is set to increase considerably, first by ~30% with the initial electrification and modernization project, and by ~60% once the system is running at 6 trains per hour with 8 cars each. (If you don't count standees, those figures are ~10% and ~25%, but why would you not count standees?) A 60% capacity increase is equivalent to one and a half lanes added to the entire length of highway 101 from San Jose to San Francisco.

It doesn't have to stop there: more trains per hour and longer trains are possible, because EMU trains scale up in a way that diesel can't. A future Caltrain capacity increase to about 10,000 passengers per peak hour per direction (about triple today's throughput) isn't out of the question, does not require adding tracks or expanding the rail corridor, and would equate to adding 5 new freeway lanes.

In certain quarters of Silicon Valley that are enamored of Hyperloops, self-driving Teslas and Boring underground tunnels, electric Caltrain is looked down upon as a last-century technology that is about to be made obsolete. That particular outlook fails to grasp the importance of throughput or to recognize the enormous carrying capacity of modern electric rail. Self-driving Teslas and Hyperloops will achieve dismal throughput capacity as measured in passengers per hour, and no amount of whiz-bang technology will change the underlying geometry of this increasingly urban region.

The way forward is to add more freeway lanes of Caltrain.

03 July 2017

The Overtake That Won't Be

In its renewed environmental review process for the San Francisco to San Jose project, the high-speed rail authority is considering the alternatives for the peninsula rail corridor. The outlines of the new draft EIR are emerging, and this is where politics meets engineering.

Interested stakeholders keep asking about how the blended system will actually work, with Caltrain and high-speed rail sharing the scarce resource that is track capacity. The issue is being studied in some detail behind closed doors by an entity known as the Joint Scheduling Working Group (JSWG), consisting of experts from HSR and Caltrain aided by their respective consultants. As of the end of 2016, the JSWG had produced a first report on its work, which was shaken loose by a public records request from CARRD. Before digging into this, let's take a look back at how we got here.

2012 Blended Operations Analysis (Caltrain/LTK)

While the original four-tracks-all-the-way HSR plan was collapsing in a firestorm of community opposition, Caltrain commissioned a study of blended system operations (8 MB PDF) from consultant LTK Engineering Services, to see if blending Caltrain and HSR primarily on the two existing tracks was viable. The 2012 study concluded that without any additional tracks, the corridor could support up to 6 Caltrain + 2 HSR trains per hour per direction, increasing to 6 Caltrain + 4 HSR with overtake tracks.  Key results were:
  • With speeds limited to 79 mph, the most reasonable option with 6 Caltrain + 4 HSR was a "short middle overtake" between Hayward Park (San Mateo) and Whipple Ave (Redwood City).
  • A "long middle overtake" all the way through Redwood City provided only marginal performance improvements.
This study legitimized the blended system, which has ever since been the favored approach to bringing HSR to the peninsula rail corridor. The study contained numerous disclaimers to the effect that no official decisions had been made regarding future service levels, programmed overtakes, stopping patterns or scheduled trip times... all the important considerations that feed into a railroad's product, namely its timetable.

2013 Additional Blended Operations Analysis (Caltrain/LTK)

A short while later, LTK published another report (2MB PDF + 14MB Appendices) that can best be characterized as an expansion and refinement of the 2012 analysis, considering additional options. Key results were:
  • Unlike the 2012 study, the "long middle overtake" performed significantly better than the "short middle overtake."
  • A new option, the "middle 3-track overtake" (between San Mateo and Palo Alto) performed almost as well in simulation, although it assumed all HSR trains entered the corridor on time, unlikely in practice.
  • Other overtake track options did not fare as well.
The disclaimers continued, with the conclusions of the study being described as "educational."

2016 Joint Schedule Working Group Report (HSR/SMA)

The key chart in the SMA study
(PDF page 62 / slide 48)
The JSWG was established in April 2016, a while after HSR had engaged the services of Swiss rail consultancy SMA. To avoid a standoff between dueling agencies and consultants, Caltrain/LTK and HSR/SMA are now comparing notes on their respective plans for the blended system. The JSWG's 2016 draft year end report (10 MB PDF) provides important context to the decisions now being made to select a "preferred alternative" for the HSR blended system EIR. This study is interesting for three reasons: it is the first blended system study led by HSR, it offers insight into the evolving ideas of the JSWG, and it is not sugarcoated because it wasn't intended for wide public distribution. Key results were:
  • The "no additional passing tracks" case is shown to support 6 Caltrain + 4 HSR per hour per direction, unlike in the LTK studies, provided that headways are tightened and Caltrain passengers don't mind sitting in a siding for 6 or 7 minutes during these overtakes.
  • The "short middle 4-track overtake" degrades Caltrain trip times, since overtakes don't naturally tend to occur there.
  • The "middle 3-track overtake" performs better than any other option, thanks to allowing bidirectional operation through almost its entire length, unlike in the LTK study where half the length of the overtake track was dedicated to each direction.
Tea Leaf Reading
The grotesque station-in-the-sky
proposed for San Carlos under
the short middle 4-track option
The constant refrain that nothing has been decided yet continues to this day, but the tea leaves are becoming quite readable. Here is some informed speculation:
  1. The HSR team really, really doesn't want to build the "short middle 4-track overtake," generally because they have no money and specifically because the SMA analysis has shown this scenario to be a poor performer operationally.
     
  2. However, the HSR team is reluctant to withdraw any alternative this late in the preparation of an EIR, after it was carefully introduced to the public through countless outreach meetings, workshops and open houses. Sudden change scares people.
     
  3. In order not to build the "short middle 4-track overtake," the HSR team has engineered it into a straw man alternative, using the prospect of grotesquely massive concrete viaducts towering fifty feet over San Carlos and Belmont to strategically elicit vigorous public opposition. It's working, but unbeknownst to them, San Carlos and Belmont have little to worry about.
     
  4. Due to having no money, the HSR team strongly favors the "no additional passing tracks" alternative. The mediocrity of the resulting Caltrain timetable, and the amount of time spent by Caltrain passengers waiting to be overtaken, is of little concern to them. But that's okay, since San Carlos and Belmont made them do it.
     
  5. The HSR team probably dreads resistance from Caltrain stakeholders who don't want the peninsula rail corridor being taken over to Caltrain's detriment. Strong resistance could force the HSR team to revive the "middle three-track" alternative that had previously been eliminated from the EIR process (see slide 15 in this outreach presentation), setting back the environmental review schedule.
     
  6. If the assumptions in the SMA analysis stand up to closer scrutiny, the "middle three-track" scenario could actually be a viable compromise for the blended system. It would no doubt be expensive due to the number of new grade separations, but the result (if one believes SMA) would be fast and robust service for both HSR and Caltrain, with reasonably-sized grade separations in every town from San Mateo to Palo Alto. In the last year, the winds of public opinion have turned more favorable to grade separations in Menlo Park and Palo Alto.
One thing is sure: the "middle three-track" alternative should be added to the HSR EIR and studied in detail, with an eye towards designing the future blended system timetable. The timetable is the product, and it will soon be time to decide on one.

07 June 2017

Frequent Trains Off Peak

After electrification, Caltrain aspires to operate off-peak service at 2 or 3 trains per hour, instead of the current 1 train per hour. All-local service at 3 trains per hour works out to a fleet requirement of 12 trains in service, far less than needed for rush hour, but still racking up almost 300 train-miles per hour, or triple today's rate. That sort of service level will not be cheap to operate, unless two conditions are met to reduce operating and maintenance costs:

1) Operate Short Trains Off Peak

Shorter trains off-peak reduce maintenance costs by putting less wear and tear on the vehicles and track. The same revenue train-miles can be offered with fewer car-miles. The more off-peak service is provided, the greater the savings: at 3 trains per hour, operating 4-car EMUs instead of full-length 8-car EMUs off-peak results in a huge reduction of 25% fewer weekday car-miles.

Operating and vehicle maintenance
costs of US commuter rail, per car mile
Just how big are the savings? Typical commuter rail costs are available from the FTA's National Transit Database. The operating and vehicle maintenance costs for Caltrain and selected commuter rail operators are shown at right for the year 2015, normalized by the total number of car-miles operated. Some on this list (Metro North, LIRR, SEPTA and New Jersey Transit) operate sizable fleets of EMUs, but their maintenance costs are not significantly out-of-family with Caltrain; therefore, it's fair to assume that maintenance costs will not materially change after electrification. Since the FTA maintenance totals are not broken out by fixed and variable costs, we will conservatively assume that the variable cost (which scales directly with the number of car-miles operated) accounts for half of the vehicle maintenance cost. Squinting at the chart, let's estimate this variable cost at $2 per car-mile.

When you operate 12 hours of off-peak service at 300 train-miles per hour, the variable cost of vehicle maintenance racks up at 12 hours/day * 300 train-miles / hour * 8 cars/train * $2/car-mile = $58k/day. By reducing off-peak train length to 4 cars/train, the savings are half of this, or $29k/day. The savings from shorter trains accrue not just on weekdays but on weekends too, yielding annual savings of roughly $10 million.

Then you might want to factor in energy cost savings. Each car weighs about 60 tons loaded, and is accelerated to about 60 mph between two typical stops. The electricity consumed to accelerate is re-generated into the grid while braking for the next stop, with a round-trip efficiency likely in the neighborhood of 80%. That means overcoming the inertia of one car for one stop (neglecting drag) takes 4 MJ of electricity, or 1.2 kWh in more familiar units. At typical electricity rates of 12 cents/kWh, that's just $0.14/car/stop. Multiplying it up, $0.14/car/stop * 20 stops * 3 trains/hour/direction * 2 directions * 12 hours/day * 8 cars/train = $1600/day.  (Note that drag will significantly increase this figure, but can be neglected for this estimate because the drag of a 4-car train is similar to that of an 8-car train.) By reducing off-peak train length to 4 cars/train, the savings are $800/day. At less than $300k per year, this is just a rounding error compared to the vehicle maintenance, and can be ignored.

The Scharfenberg automatic coupler,
nicknamed "Schaku," linking up two
short EMUs (click for movie)
Offsetting these savings are the costs of making and breaking train formations several times per day, since the entire fleet needs to be available for morning and evening peak service with full length 8-car EMUs. Traditionally, this is a cumbersome operation that involves expensive and specialized labor, with ground crews stepping onto the tracks to connect pneumatic hoses and high-voltage cables. Caltrain is breaking with tradition by using a neat technological trick: the couplers on each end of the new EMUs are fully automatic Schakus, making mechanical, pneumatic and electrical connections in a matter of seconds at the touch of a button in the train cab. Barring any union rules relating to craft distinctions, making and breaking trains can be performed by train crews with zero additional labor cost.

2) Operate With One Conductor

Labor accounts for about two thirds of operating costs in typical commuter rail systems. Operating costs are strongly driven by train crew size. Minimum crew size is constrained by union rules that govern how many conductors must work on each train. Currently, the minimum crew size (dictated by Rule 11 of the agreement with the UTU) is 1 engineer, 1 conductor and 1 assistant conductor for trains up to seven cars, with a second assistant conductor required for an 8-car train or longer.

When contemplating a tripling of off-peak service, the cost of this minimum staffing level becomes prohibitive. Conductors are paid about $40/hour, and assistant conductors about $35/hour. Including benefits and other employee costs, the overall cost of these employees is easily double these figures. Additionally, conductors typically spend about half their shift time on board a revenue-producing train, so the necessary staffing levels are roughly double the number of trains in service. We saw earlier that it takes a fleet of 12 trains to operate off-peak service at 3 trains per hour per direction; staffing an assistant conductor on these trains would cost $70/hour/conductor * 1 conductor/train * 2 hours/(revenue hour) * 12 trains * 12 (revenue hours)/day = $20k/day. Again this is big money: the savings from removing the assistant conductor and going to one-conductor operation accrue not just on off-peak weekdays but on weekends too, yielding annual savings of roughly $7 million.

How do you sell this lower staffing level to the union?
  1. EMUs can relieve conductors of some of their workload, after automation of many of their traditional roles (such as stop announcements, door and lift operation, or signal aspect acknowledgement). Fare verification (proof of payment) could even become a separate role carried out by roving fare inspectors.
  2. Conductor staffing levels or pay rates can be renegotiated on the basis of actual ridership, instead of the number of train cars, since the new EMUs will have automatic passenger counters that collect detailed and accurate passenger ridership statistics.
  3. Most importantly, the total amount of work for UTU-represented employees would increase, since one-conductor operation would enable a tripling of off-peak service, resulting in 1.5 times more labor hours even after cutting conductors staffing levels in half.
It isn't a stretch to envision Caltrain and the UTU re-negotiating the labor agreement to allow just one conductor on four-car off-peak trains; there is room for a compromise that can benefit everyone.

Future Fleet Implications

If you zoomed way, way, into Caltrain's
exterior paint scheme concepts,
the Schaku was plain to see
Caltrain's initial fleet of sixteen six-car EMUs (total 96 cars) will not have the ability to split into shorter formations, but once the option for 96 additional cars (total 192 cars) is exercised, and all trains are extended to their intended length of eight cars, the practice becomes not only possible, but necessary for providing frequent off-peak service.

The fleet needs to operate two service patterns:
  1. peaks at 6 trains per hour with a fleet of 8-car EMUs
  2. off-peak at 3 trains per hour with a fleet of 4-car EMUs
To support both service patterns using the planned fleet size of 192 cars (including a rather large spares ratio, to withstand regular grade crossing collisions), the optimal fleet configuration is probably something close to:
  • 16 4-car EMUs for off-peak service, each with one bike car and one bathroom car, that can be coupled in pairs during peak hour service to form eight trains with eight cars each.
  • 16 8-car EMUs for peak service, lengthened from the base order
This results in the following order breakdown for the 96 additional option cars:
  • 32 passenger cars for CalMod 1.1
  • 32 cab cars, for 4-car EMUs
  • 16 bathroom cars (powered), for 4-car EMUs
  • 16 bike cars (unpowered), for 4-car EMUs
This EMU fleet configuration enables 20-minute off-peak service frequency for at least $17 million/year cheaper operating and maintenance cost than would otherwise be achieved with a uniform fleet of all 8-car trains. That's a large amount, easily over 10% of Caltrain's current annual operating budget. Considering that Caltrain struggles every year to scrape together enough operating funds, a stronger way of stating it is that without 4-car EMUs and one-conductor train crews, Caltrain will simply not have the financial means to provide 20-minute off-peak service frequency.

27 May 2017

CalMod 1.1

This being Silicon Valley, future plans for Caltrain modernization are known as CalMod 2.0, the next big thing beyond the CalMod 1.0 improvements that are already under contract.

CalMod 2.0 is a list of future improvements worth about $750M that includes:
  • Full fleet conversion to 8-car EMUs ($440M)
  • Broadband connectivity ($30M)
  • Maintenance facility improvements ($36M)
  • Level boarding and platform extensions ($250M)
In the grand scheme of things, these aren't outrageous expenses ("only" another 38% over and above the $2B tab for CalMod 1.0), but they're not cheap, either. To meet capacity challenges in the short term, possibly concurrently with delivery of CalMod 1.0, perhaps some of these expenses can be moved up to realize the maximum bang for the buck as soon as 2021.

CalMod 1.1 would consist of just two line items:

1) Lengthen EMUs to 8 cars, for $145M

The EMU fleet for CalMod 1.0 consists of sixteen 6-car trains, with a reduced seating capacity of 558 that has caused much yammering amid the increasing load factors during peak commute hours. Even without a ridership bump from the "new and modern" effect, it is likely these trains will be packed from day one. Now is the time to start doing something about it.

Seating layout for two extra cars (based on Stadler brochure)
Two unpowered cars would seat up to 264 passengers.
The contract with Stadler includes an option for another 96 cars priced at $390M, a figure larded up to $440M in the CalMod 2.0 total presumably due to the usual procurement overheads. This figure is for 100% fleet replacement, with all the remaining diesel consists being retired. In the short term, only 1/3rd of the option cars would need to be exercised; this involves purchasing 32 cars or 2 extra cars for each of the sixteen EMU consists in the CalMod 1.0 order.

The per-train capacity will increase by well over 200 seats per train, back to a level that will mitigate peak hour crowding. However, 8-car EMUs will exceed the length of many of the existing platforms.

2) Extend platforms to a minimum length of 700 feet, for $25M

Platform extensions are relatively cheap to build, especially when you don't need to rebuild the entire length of station platforms as would be needed for level boarding. You can leave vertical circulation (stairs, ramps) and amenities (vending machines, lighting, benches, PA system, departure boards, etc.) alone and just tack on a short length of concrete, and perhaps move a pedestrian crossing. Caltrain excels at building platforms and has done so extensively, pouring some 5 linear miles of platforms over the last 18 years!

The length of Caltrain's existing platforms is documented in this schematic of California rail systems. To dock an 8-car EMU, platforms need to be extended to at least 700 feet. The necessary extension lengths are graphed at right; labels show the year of completion of each platform's construction.

The total amount of platform extension required to operate 8-car EMUs is approximately 3500 feet. This figure excludes Hillsdale and South San Francisco, both of which are already slated to be rebuilt to 700 feet. Each foot of platform costs about $7000 to build, on the basis of a typical $10M cost for two 700-foot platforms from past platform reconstruction projects. Therefore, the tab for extending all platforms to 700 feet (for the time being, at their current height of 8 inches) lies in the range of $25M.

Start Planning Now

The bottom line: another $175M or an extra 9% investment over CalMod 1.0 yields an extra 23% peak hour seated capacity for CalMod 1.1. It would be best to start planning for CalMod 1.1 now, and to turn CalMod 2.0 into the big level boarding project for the 2020s. In software parlance, the CalMod 1.1 patch should be applied immediately upon release of CalMod 1.0.