25 September 2019

Risk and Opportunity in Redwood City

Lowe, a major real estate development firm, is preparing to redevelop Redwood City's Sequoia Station, an outdated strip mall adjacent to the Caltrain station, into a 12-acre mixed-use project with towers up to 17 stories tall.  If that is eye-opening to residents of Redwood City, consider that few people yet know that a greatly expanded Redwood City station is the keystone transfer node to enable the growth envisioned in Caltrain's business plan service vision. This new station will require slightly more land than the railroad already owns, and can only be located in Redwood City, the sweet spot that lies halfway between San Francisco and San Jose at the connection point to the Dumbarton rail corridor.

This creates a risk: if a commercial development project is allowed to proceed without respect to the future real estate needs of the railroad, then Caltrain will be constricted and unable to build the optimal infrastructure to support future growth.

Additional Land Needed For Caltrain

Caltrain and Samtrans have extensive land holdings at the Redwood City transit center. Still, just a bit more is needed to build a high-functioning piece of infrastructure, and be could traded for other parcels. Click to expand the map:

Land needed for future expanded station in Redwood City (shaded green)
Design Principles

The absolute worst way to build it.
Existence of this city rendering is
reason enough to be concerned.
To ensure that the Sequoia Station project becomes an exemplar transit-oriented development, rather than relegating Caltrain to the role of development-oriented transit, the rail agency and the developer should agree on some broad design principles.
  • Think Big. Redwood City is one of the few stops on the peninsula rail corridor not surrounded by a sea of low-density single-family housing. Intensive land use and transportation must fit together to achieve a dynamic yet sustainable low-carbon future.
     
  • Form follows function. No amount of architectural flourish or amenity can make up for a poor station design. Optimize for convenient access, easy transfers between trains and buses, short walks, direct and intuitive routes.
     
  • Put the station at the center of the action, right over Broadway. Don't shove it to the north, out of the way of the development. The city rendering at right shows precisely what NOT to do.
     
  • Configure the station as two island platforms to facilitate cross-platform transfers, without time-consuming vertical circulation or platform changes. The Caltrain business plan's staff-recommended service vision relies entirely on these Redwood City cross-platform transfers; every single train that pulls into Redwood City will make a timed transfer to another same-direction train docked at the opposite edge of the same platform. Denoting express tracks as 'F' for Fast and local tracks as 'S' for Slow, the optimal layout is FSSF with two islands, resulting in F-platform-SS-platform-F. Again, the city rendering shows precisely what NOT to do: passengers would not only have to change platforms, but also cross the tracks at grade.
     
  • Elevate the train station to reconnect the street grid and make the railroad permeable to pedestrians, bikes, and other traffic. A busy four-track station is fundamentally incompatible with at-grade railroad crossings, and the only reasonable way to grade separate at this location is by elevating the entire station. Obstacles to pedestrian circulation such as the Jefferson Avenue underpass would be removed. Once again, the at-grade city rendering shows what NOT to do.
     
  • Use four-track approaches from the north and the south. Cross-platform transfers are most efficient if trains do not have to arrive and depart sequentially using the same track, which adds about 3 minutes of delay. The best transfer is one where the two same-direction trains can arrive and depart simultaneously on their own separate tracks. Temporal separation is efficiently established by having the local train stop one station away from Redwood City (southbound at San Carlos or northbound at a new Fair Oaks station at Fifth Avenue) at each end of a new four-track segment that will ultimately measure four miles. In this arrangement, the express trains naturally gain on the local trains without a single passenger being delayed at Redwood City.
     
  • Include turn-back tracks. Preserve room in the right of way north and south of the station for turn back pocket sidings, between the central slow tracks. Dumbarton rail corridor trains may not necessarily "interline" or continue on the peninsula rail corridor, so it's important to give them a convenient place to transfer and turn around without fouling other train traffic on the express tracks (hence FSSF arrangement). Same thing for a possible San Mateo local, which could serve the more densely spaced stops north of Redwood City.
     
  • Don't be constrained by discrete city blocks. It could make sense to build structures or connect them over and across the tracks, more tightly knitting the station complex into surrounding mixed-use neighborhoods. This has some surmountable safety and liability implications, but buildings on top of busy stations are a common feature of successful cities around the world.
     
  • Plan for long 400-meter platforms, not Caltrain's standard 700-foot platform length (again as seen in the city rendering of what NOT to do). While statewide high-speed rail plans currently do not include a stop in Redwood City, it is becoming enough of a destination and a regional transportation node that it makes sense to build a station large enough to future-proof it for service by long high-speed trains, regardless of what the California High-Speed Rail Authority might have to say about it.
     
  • Think ahead about construction sequencing. Redwood City should be grade separated in one project from Whipple to Route 84, including the elevated station, taking advantage of Caltrain's land holdings to minimize the use of temporary tracks. A shoo-fly track would have to be built on Pennsylvania Avenue (within the railroad right of way) to make room for construction of the western two-track viaduct. Trains would begin using the elevated station while a second eastern two-track viaduct is constructed. Pennsylvania Avenue could re-open later, under the new four-track viaduct. Construction sequencing may drive how much extra land is needed for the railroad, so it's important to think it through up front.
If these design principles are respected, the re-development of Sequoia Station will present not a risk but an amazing opportunity to enhance Redwood City by realizing its full potential as the fulcrum of the Caltrain corridor and of a new regional express network reaching across the Dumbarton bridge and beyond.

01 September 2019

Electrification Delayed

Caltrain's electrification project is showing ominous signs of falling badly behind schedule. There are at least five bearish indicators:

Slippery milestone
Slipping Milestones. One key milestone reported in the project's monthly progress reports is known as "Electrification Substantial Completion." From the December 2018 report to the July 2019 report (over a span of 7 months), the milestone has slipped from 6/23/2021 to 12/31/2021 (a bit over 6 months). When a major milestone slips almost day for day, you know the project has gone sideways. The latest PMOC report from the FTA shows that the contractor's date for this key milestone has slipped well into 2022, over a thousand days late relative to the milestone date promised when the contract was signed.

Severely under spend plan
Significant Under-Spending. The amount of money spent to date is about $640 million less than planned at the start of the project. If the value of the work accomplished is commensurate with the amount spent, then the project is 1.5 years behind schedule. However, there are strong indications of inefficiencies (such as "differing site conditions" disrupting foundation installation) and unplanned scope (such as the new grade crossing constant warning time solution) that make it exceedingly likely that the value earned so far is less than had been planned for the amount spent. From an earned value perspective, the CPI is likely under 1 (over budget) and the SPI below 0.6 (further behind schedule than the spend curve might imply).

The little engine that couldn't
Foundation Chaos. As is plainly obvious to anyone riding the train, foundation installation is not a spatially or temporally orderly process. Digging into the ground reveals old utilities, and often reveals the recently-installed CBOSS fiber optic cables, evidently placed by the contractor where it was easiest (right where foundations need to go) with the as-built configuration either incorrectly documented or not at all. This is another CBOSS issue that could end up in court. Conflict with these cables does not bode well for PTC testing or activation, or for the cost of foundation and pole redesign and relocation. Recent indicators show a slight uptick of foundation productivity, but it still lags well below the monthly average of 174 that must now be sustained every month to complete on time. The all-time record is 122, and indications are that August 2019 totals have slid back considerably below trend.

Missing tasks are delayed and
on the critical path
Missing Schedule Tasks. By all accounts, the long pole in the tent (the critical path of the Balfour Beatty schedule) is the design, installation and testing of the signal system modifications, including the new grade crossing warning system. However, such a task is nowhere to be found in the schedule published in Appendix C of the monthly report, which obscures any insight into the true status of the project. Having recently set $150 million on fire with CBOSS, Caltrain is understandably skittish about revealing further unforeseen costs and delays associated with signalling, but it seems inexcusable at this juncture that the public master schedule would show only "OCS," "Traction Power," and "Segment Testing" tasks for the electrification contract, when all the action is in the missing task "Signal System Modifications," which is very much on the critical path in Caltrain's internal schedule and the contractor's schedule.

Proliferation of Schedules. There is apparently no agreement between Caltrain and their contractor on what the real program schedule is. The public schedule in the monthly report is served with a cautionary statement that Balfour Beatty is reporting a significant delay, but the completion milestone is still optimistically set to 12/31/2021. When you end up with several schedules, there is effectively no longer a project schedule. It's anyone's guess when the project will be done, and chances are increasing rapidly that it won't be in 2022, despite Caltrain's increasingly desperate insistence that everything is fine.

Right now would be a good time to come clean about what's really going on. Total transparency is the only saving grace that can spare Caltrain from accusations of project management incompetence.

28 July 2019

Emergency Exit Fail

Caltrain's new EMU train cars have an unusual configuration with two sets of doors. The lower level doors will be used at existing Caltrain stations, while the intermediate level doors (above the wheels at the ends of each car) are intended to be used at an undetermined date in the 2030s once these trains begin sharing stations with California high-speed rail, which will use high-floor trains and high platforms with boarding at about 50" above the rail. The California High-Speed Rail Authority, which Caltrain cryptically refers to as "external stakeholders," required this design feature as a condition of funding Caltrain's modernization to the tune of $750M, to maintain the option of sharing platforms at future HSR stations in San Francisco, Millbrae and San Jose.

The Original Plan

To maximize the short-term seating capacity of the new trains until the 2030s, Caltrain specified that the intermediate level should have temporary flip-up seats installed in front of the unused doors, five per door vestibule, with the seating blocking off the doors like this:
Configuration of intermediate level in A, B, C, E, and G cars
Because EMU cars are filled with electrical cabinets (labeled with yellow lightning bolts), the seating capacity of the train is reduced compared to a conventional train. This is the price you pay for not having a locomotive; all the bits that make the train go still need to find a place, which makes for a challenging packaging problem in a bi-level train. The reduced seating capacity of the train has been controversial and makes these temporary seats quite important. For each 7-car train, there are 70 of these intermediate level flip-up seats that make up a non-trivial 10% of the overall seating capacity of 667.

At some undetermined future date when the intermediate doors would be needed for compatibility with high platforms, the blue flip-up seating modules would be removed from the intermediate level.

A Regulatory Conundrum

In the design of any new train, federal safety regulations require that any passenger seating compartment be fitted with at least two emergency exit windows (for passenger egress) and two rescue access windows (for first responder ingress). The intermediate level counts as a passenger compartment because these flip-up seats are located within it. However, the intermediate level does not have what regulations consider to be a window; the only opening to the outside is through the doors. This set up a conflict with safety regulations.

In late 2017, Caltrain petitioned the Federal Railroad Administration for a waiver (docket FRA-2018-0003) by arguing that the emergency release feature of the doors would provide an equivalent level of safety, despite not meeting the letter of the regulation, allowing emergency access by climbing over the seat backs.

In June 2018, the FRA denied Caltrain's request because the flip-up seating installed longitudinally such that it blocks the doors could impede egress and access and therefore did not meet the intent of the regulation. The FRA stated that "the absence of need for these intermediate level doors to support current revenue boarding and alighting requirements does not negate the necessity for an unobstructed path in the event of an emergency." Curiously, this unobstructed path requirement applies only to doors, not to windows!

Implicitly, Solution A is to remove all seating from the intermediate level of the affected cars, which effectively sidesteps the emergency window requirement. But given that seating in Caltrain's EMUs is already quite limited, this solution seems like a non-starter as it would reduce seating capacity of a 7-car train by 9% from 667 seats to just 617 seats.
Solution A: not a passenger seating compartment
The FRA helpfully suggested some other possibilities.

Solution B: equip the intermediate level doors with a regulation-size emergency window of minimum dimensions 26" wide by 24" high. Unfortunately, that is too large for the dual-leaf design of the train doors; in other words, the window in each door leaf is too narrow to function as an emergency window.
Solution B: the minimum clear opening is too big for dual-leaf doors
Solution C: replace the intermediate level doors with a plug panel (essentially, a structural wall panel that does not function as a door) fitted with a regulation-size emergency window of minimum dimensions 26" wide by 24" high, until such time as the door-blocking seating is removed, the panel is removed, and the doors and platform bridge plates are re-installed.

Solution C: doors replaced by plug panels
Caltrain is now in the process of pursuing Solution C, plug panels. This change order is expected to cost about $4 million total up front, about $30000 per car, or $7000 per door. When intermediate-level doors are required a decade or more from now, a net sum of approximately another $10 million ($14 million future installation cost to be set aside, minus $4 million of door maintenance savings) would be needed to retrofit them. That is a LOT of money for a change that fundamentally reduces and complicates compatibility with HSR stations and platforms.

Other Solutions

There are other solutions that strike a better balance of functionality and simplicity without a seven-figure cost impact.

Solution D: short of removing all the seating from the intermediate level vestibule, the regulations require only one emergency window (instead of two) if there are four or fewer seats in the compartment. Removing seats from one side only and applying for a new waiver to allow unobstructed use of one of the doors in lieu of a single emergency window could work, addressing the FRA's stated concern with door obstruction. This would reduce seating capacity of a 7-car train by just 22 seats or 3% (5 seats lost in cars A and B, and 4 seats lost in cars C, E and G).
Solution D: reduced seating with unobstructed emergency access
Solution E: reconfigure the mounting bracket for the flip-up seating so that seats flip up and out of the way of the doors when not used, allowing the unimpeded use of both doors in lieu of emergency windows. This solution requires applying for a new waiver to allow the use of doors in lieu of emergency windows, but also addresses the FRA's stated concern with door obstruction. Placing the flip up seats in this manner would reduce the clear width of the door opening by a couple of inches on each side, from 51" to about 47", with no reduction to seating capacity.
Solution E: change flip-up seating orientation to provide unobstructed door access
(flip-up seats are shown in use; they fold flush against wall when not occupied)
Solution E would require no modifications whatsoever when the intermediate level doors are needed in the future, and could be implemented at all doors throughout the train including the lower level, adding seating capacity. Seats placed in doorways may sound like a bad idea, but in a crowded train, social signaling fairly quickly communicates to occupants of these seats that it's time to stand up and make way. This is the French "strapontin" seating in common use on some of the busiest rail lines in Paris:

Flip-up seats in a doorway of a brand new Bombardier EMU on Paris RER line D.
(foreground at left) credit: Wikipedia / KiHa 52
Indeed, the photo above, taken inside the same Bombardier EMU often vaunted in front of the Caltrain board by a certain member of the public as having so much more seating than Stadler's EMU, shows one of the secrets of achieving very high seating densities: flip-up seating in all doorways. The other three secrets are five-abreast seating, not having as much space dedicated to bikes, and lower acceleration performance that requires fewer electrical cabinets, leaving more space for seats. After adjusting for these four factors, it turns out that the Bombardier EMU provides no higher seating density than the Stadler EMU.

Ultimately, it is entirely possible that Caltrain simply does not wish to interface with high-speed rail in any station as a matter of policy, because it would require sharing and collaborating with another agency, and solving a somewhat complicated ADA compliance problem. Which agency would voluntarily bring that upon itself? Caltrain already took the HSR money, and installing plugs will "erase" the clunky and unpalatable concession they made in the name of compatibility, with the further bonus of not requiring another run at the FRA for a new waiver. The complicated ADA compliance issues associated with interior lifts are kicked as far down the road as possible!

No matter how you look at it, Caltrain's chosen approach is a ~$15 million mistake that reduces and complicates compatibility with HSR stations and platforms. There are cheaper, simpler and easier ways to achieve compliance with emergency window regulations. It's not too late to change course.

05 May 2019

Thoughts on Grade Separations

The emerging Caltrain business plan is broaching the issue of grade separations, a decadal process that has been underway, well, for decades. We're already 63% of the way there today, with another dozen new grade separation projects in various stages of planning or construction. Achieving a reasonable level of grade separation for the peninsula corridor is estimated to cost $8.5 - 11.1 billion, a shockingly large sum that we'll just round to $10 billion. As we try to grasp the enormity of that figure, here are some contrarian thoughts:

1) Don't spend train money on car projects. The benefit of grade separations accrues primarily to automobile travel, with the elimination of gate down time. An intensive grade separation program can eventually unlock additional operating slots for more trains and eliminate the occasional incident, yielding benefits to train riders. Some grade separations are necessary, such as when expanding to four tracks. In the short term, however, the greatest benefit is the removal of an inconvenience to drivers, which in our car-centric society is held as a worthy goal seemingly regardless of cost. Rail dollars are a lot scarcer than road dollars, especially in this era of federal disengagement, so the last project we should spend them on is a project that facilitates car travel with little improvement for train riders. Rail funding should be used to make real and measurable improvements to train service, a standard by which most grade separations rate poorly. So you still want a grade separation? Build it with road funding.

Anticipated gate down times,
under various scenarios in the
Caltrain business plan
2) Quit whining about gate down time. Caltrain put together a nice summary of gate down time, the number of minutes per hour that grade crossing gates block traffic during rush hours. Today the average is 11 minutes, and under future growth scenarios it could increase to 17 - 25 minutes, with a few crossings faring worse than average. If that sounds intolerable, think about a typical roadway intersection with a traffic light. If both roads are equally important, the "gate down time" of a traffic light is 30 minutes. If one road is more important, the lesser road (for example, Ravenswood Ave where it meets El Camino Real in Menlo Park) sees "gate down time" well in excess of 30 minutes, let's say 40 minutes per hour. Nobody is clamoring to grade separate the Ravenswood / El Camino road intersection. There's an obvious double standard here, and the guidelines for what qualifies as unacceptable delay should be set the same way as they are for the grade separation of a road intersection. Gate down time should only rarely, if ever, be the reason to build a new grade separation.

3) There are few economies of scale in grade separation. Doing them all as a package does not save money. The process we have, where local jurisdictions often exert tight control over every aspect of design and construction, does not lend itself to a one-size-fits-all approach. Each grade separation is different. Grade separation designs do not depend on each other in the majority of cases where they are widely spaced. While a corridor-wide strategy is important to have, the execution of that strategy and the securing of funding is inherently a city and county issue. If we are going to have a corridor-wide funding approach, it must go hand-in-hand with taking away local control. Jurisdictions that insist on local control should be left to figure out the funding on their own. Palo Alto, where interminable and futile discussions of tunnels continue to this day, should not be allowed to control the design process if their project is paid for through a corridor-wide funding measure.

4) If $10 billion is an okay expense, then there are far better ways to spend it. Especially with rail money at stake, there are much better ways to spend $10 billion than by building a lot of grade separations for cars that produce zero improvement to train service. There are a lot of good investments that should be made to improve the amount and speed of train service:
  • Extend all platforms to 8-car length. If you put all the platforms that Caltrain ever built in the last 20 years end to end, they would stretch about 5 miles long. This is not an expensive project; it can be done for about $0.05 billion. It should already be underway, but inexplicably isn't.
  • Convert the entire train fleet to 8-car EMUs, starting by exercising the rest of the existing Stadler contract option of another 59 cars, increasing the fleet to 24 trains. The diesels are retired from the peninsula, which is a condition for starting any level boarding projects. This costs about $0.4 billion.
  • Convert the entire system to level boarding to speed trips and improve punctuality. Depending on how this is done (high platforms or low platforms, or some combination thereof) and over how long a period of construction, this would cost about $0.5 - 1 billion.
  • Build a new EMU maintenance and storage facility near Blossom Hill (San Jose) and extend frequent electrified service through all of San Jose. Including any extortion by UPRR, the owner of the tracks, this ought to be feasible for less than $1 billion.
  • Build a new transit center in Redwood City to enable cross-platform transfers between locals and expresses. Call it $0.5 billion, and throw in the downtown grade separations for another $0.5 billion to allow four tracks.
  • Expand the EMU fleet to enable 8 train per hour peak service. Expanding the fleet to 32 trains would require another 64 EMU cars, for about $0.5 billion.
  • Extend the platforms at highly patronized express stops to 12 cars in length, and extend expresses to 12 cars. This would require extending about half the fleet by 4 cars, or another 64 EMU cars. Including platforms this might cost about $0.8 billion.
This isn't an exhaustive list, but unlike grade separations, all of these projects have immediate and measurable positive effects on the quantity and quality of service provided to riders. This list achieves most of Caltrain's "moderate growth" scenario but without HSR. The tally for all of these projects is still less than $5 billion, so if $10 billion for grade separations sounds at all palatable, this list ought to be a no-brainer.

Grade separations are nice, but their cost and benefit should be weighed very carefully on a case-by-case basis. The cost should be borne by who benefits. The business plan process will hopefully create the framework to have the difficult conversations about what not to pay for with rail funding. Grade separations should be built with highway funding unless there is a clear and measurable benefit to rail service.

24 April 2019

Foundation Progress Tracker

One way to measure the progress of a large and complex construction program like the Peninsula Corridor Electrification Program is to count how many foundations have been completed. This is a revealing metric, since foundation construction is currently the top risk on the program due to surprises when digging holes along the right of way. It's also a metric that is readily measurable and reported monthly.

In round numbers, the electrification project encompasses ~2500 poles and ~3100 concrete foundations. The number of foundations is greater than the number of poles because there are foundations for guy wires and sometimes multiple foundations for portal poles.

The progress chart below will be updated monthly.


At the December 2018 meeting of the Caltrain board of directors, the program manager stated (starting at 01:03:00 in video) that he needed to maintain a pace of 156 pole foundations per month (six per night) to meet the schedule milestone of "electrification substantial completion," which was then set for June 2021. You can see how things went since then.