BART to Silicon Valley Phase 2 – time to prune costs

As the lockdown response needed to save lives from the pandemic opens deep holes in public sector budgets, the escalated costs of the next phase of the BART Silicon Valley project are looking ever more unattainable and unwise. Two factors are causing costs to  climb.

First, an unprecedented tunneling method, proposed to reduce construction disruptions to Downtown San Jose, is technically complicated and costly to build. Second, extending passenger service north of Diridon Station to Santa Clara duplicates Caltrain, providing minimal benefits to riders at enormous added cost. 

The VTA board will be holding a study session on Friday, April 17 staring at 9am to consider the BART Phase 2 project – it is time to reconsider tunneling method and route alignment choices to reduce costs and construction risks to ultimately deliver a successful project in a reasonable timeframe. 

Last fall, VTA disclosed that Bart Silicon Valley Phase 2, which will extend the line from Berryessa to Downtown San Jose, had been delayed by four years to 2030, and the estimated cost had risen by $800m to $5.6bn (Old Budget; New Budget)   These time and cost increases go well beyond the delays and problems with the Silicon Valley BART project’s Phase 1 extension to Berryessa, with opening recently pushed back to June 2020, four years later than the originally promised 2016 opening.

We support the BART Phase 2 extension and celebrate its high-level goals: connecting Downtown San Jose, finally closing the regional rail ring around the Bay with a connection at Diridon station, and supporting infill development. 

You can share your thoughts with the VTA board, by calling into the teleconference board workshop starting at 9am and make an audio public comment – see the agenda and links to instructions here.   Or send a note to the VTA board secretary, board.secretary@vta.org.

Opportunities to reduce costs

As design for Phase 2 is only two percent complete, we can expect further delay and cost increases as design and construction progress. As the Covid-19 recession deepens financial challenges, decision-makers have the opportunity and the obligation to look at opportunities to prune costs, including. 

  • Reconsider the decision to require wide bore tunnels to accommodate side-by-side tracks, which has added $800 million to the project cost
  • Revisit the twin-bore tunnel option to reduce cost and risk
  • Revisit options to reduce the amount of tunneling, including use of cut and cover construction for stations with methods to reduce street disruptions and business impacts
  • Review station designs, considering more compact designs without mezzanines to reduce costs and make passenger access faster and easier. 
  • Eliminate the redundant segment between Diridon and Santa Clara, which parallels Caltrain
  • Support initiatives for more coordinated regional transit planning and project delivery, to reduce these sorts of challenges in the future.

Wide-Bore Tunnels

In 2017, VTA and BART announced that they were investigating the use of a single-bore tunnel for Phase 2. While the existing BART system was dug as two small (20ft diameter) tunnels for the two sets of rails with station platforms in between, the wide-bore approach would involve digging one large (45ft diameter) tunnel which could contain both sets of rails and the station platforms within its diameter.  Then, in the next iteration of design planning, the tunnel width was proposed to be increased from 45ft to 55ft, to accommodate side-by-side platforms in a single tunnel – using among the largest tunnel boring machines ever built – a choice that we now know has added $800m to the cost of the project.


VTA image demonstrating twin-bore vs single-bore tunnels

Single-bore construction has proved effective in some international projects, such as the Barcelona Metro. As the image below shows, in order to fit both sets of rails plus platforms into the tunnel, trains would have to run in a stacked profile (one on top of the other) rather than side by side, as in most of the BART system. While BART mostly uses side-by-side platforms, the downtown SF stations are stacked, with Muni tracks on the upper level and BART on the lower level, and the 12th and 19th street BART stations have two levels. https://en.wikipedia.org/wiki/De_Pijp_metro_station 

Comparison of side-by-side and stacked track configurations

The stacked profile has been used in other metro systems around the world, such as recently in Amsterdam, with no safety or operational issues.  Despite this fact, BART staff expressed concern that the stacked track configuration would not meet its fire safety requirements. In particular, BART expressed the opinion that evacuation procedures might be “unintuitive” for passengers. In order to allay this fear, VTA and BART convened a panel of transit experts from peer agencies such as WMATA, LA Metro, SFMTA, New York’s NYCT and MTA, and MARTA, to review the evidence and provide feedback on the approach. The result of this review was startling: even though the panel found that “the single-bore tunnel can be operated safely as an extension of the BART system”, they recommended that VTA move to side-by-side tracks in order to avoid additional training for train operators and first responders. 

As noted above, to accommodate side-by-side platforms with a single bore, the tunnels would need to be widened from 45ft to 55ft, leading to the fall announcement of an $800 million increase in cost and for additional years to deliver the project. 

We understand that training for operating staff and wayfinding for passengers for disaster situations is life-saving – especially since disasters are rare, familiar designs and standards are better. But we strongly question whether the reduced need for extra training and design is worth the cost. $800m represents an entire year of BART’s operational budget or about a year and a half of VTA’s local transit operating budget

Revisit tunnel design and construction methods

Given the budget challenges facing the region with the Covid-19 crisis, it may be worth stepping back further to look again at cheaper tunnelling approaches such as cut-and-cover or the Sequential Excavation Method.

The “sequential excavation” method has been recently used for the Los Angeles Regional Connector project where the tunnel was delivered on time . The Tunnel Technical Studies Report notes that the design aims to “reduce cut-and-cover construction” due to street-level impacts, but thanks to modern advances cut-and-cover can be employed while keeping major roads open, using the Milan Method, in which the decking is installed at the start of construction and excavation occurs below the decking (see image below). This method also has the advantage of bringing stations closer to the surface, reducing costs and making ingress and egress faster and easier for passengers.

In any event, VTA should consult with tunnelling experts from places that have successfully built metro systems at low cost. Countries including Korea, Spain, Italy, and Norway and others have a great deal of experience in this area, and it would be wise to seek guidance from any agencies willing to assist.

Lower Cost Stations

While the Supplemental Environmental Impact Statement  Financial chapter provides little detail, based on other projects it is likely that stations are contributing significantly to the high relative cost of Phase 2

As a result, stations above all should be built using cut-and-cover methods wherever possible.   Construction is isolated to just the station box area, not the entire tunnel. As an example the Eglinton Crosstown line in central Toronto is building 15 underground stations as cut and cover (with twin bore tunnels), a neighborhood that is denser than downtown San Jose, along narrower, 80 feet right of way compared to Santa Clara Street’s width of 95-100 feet.

Stations should also be limited to the smallest size necessary. Large mezzanines and open spaces are often popular with designers as they can create an impressive effect, but should be avoided as they drive up costs substantially, and can increase time for travelers to access the station platforms, and make stations more difficult to patrol and maintain. Glen Park, Balboa Park, South San Francisco, and San Bruno BART stations are examples of BART stations without mezzanines where the street level functions as the pedestrian circulation area

Eliminate redundant Santa Clara passenger service

Fragmented transit agencies are a cause of another issue with the Phase 2 project: the duplication of Caltrain services between Diridon and Santa Clara station with parallel BART tracks. As the map below shows, a substantial proportion of the tunnel, and one out of four of the stations in Phase 2, are to link Diridon station with Santa Clara. This might be a sensible extension if it weren’t for the fact that there is already a rail service between these points: Caltrain.

Image result for vta bart phase 2 2030

Phase 2 Project Map

Caltrain’s program of electrification, and the Business Plan Service Vision with goals for even greater frequency increases will mean trains running between Diridon and Santa Clara at least every 15 minutes before the BART Phase 2 extension is complete, obviating the need for any duplicate service. 

As with fire safety, good operational planning can remove any downsides. Timed transfers could allow passengers to move between BART and Caltrain services quickly and easily, adding only a few minutes to a journey. 

The BART-SV Phase 2 project includes a maintenance yard at Newhall in Santa Clara, which has been used as a rationale for building a new station in Santa Clara.  If BART continued to use the maintenance yard in Hayward, it would need to back trains up about 25 miles from the end of the line at Diridon. This is not unheard of elsewhere in the world. For example, it is common in the UK to have yards well away from the ends of the line, such as Thameslink which has its southern yard 20 miles from its southern terminus at Brighton.  The cost of the deadheading will surely be lower than the cost additional rail segment. Another option could be to revisit an earlier idea to store and maintain trains at the Las Plumas site near the Berryessa station, which would reduce the amount of deadheading to about 4 miles (see page 32 onward, here:https://www.vta.org/sites/default/files/2019-10/Chapter%202%20Alternatives.pdf

In any event, the location of a yard does not justify building a redundant station, staffing the redundant station and running duplicate passenger service.  Currently, BART occasionally stops at the Hayward Yard for staff, but but there isn’t a passenger station just because there is a maintenance yard.

Removing this unneeded station would likely save hundreds of millions, and additional millions in operating costs over time – at a time when VTA is having to prune local service because of lack of operating funding. 

Proposed Caltrain Electrified Service Plan

The original planning for the BART Silicon Valley project was done in the 90s when the Peninsula Corridor Joint Powers Board had recently rescued the commuter rail service from closure and viability was still in doubt.  A rationale at the time was that extending BART to Santa Clara would create a good position to further extend BART up the Peninsula Corridor to connect to Millbrae.  

This vision remains in Santa Clara Clara County, where VTA recently submitted a project to build BART between Millbrae and Santa Clara, parallel to Caltrain, to MTC’s PlanBayArea – where it scored poorly.  

Conditions have changed dramatically since BART SV was funded in 2000.  Caltrain ridership skyrocketed, and a $2B electrification project is under way. Even the most recent  environmental studies for BART SV were done before Caltrain electrification funding was confirmed. The BART Silicon Valley project should coordinate rather than compete with the upcoming Caltrain service increases , and eliminate the cost of the Santa Clara station, and prioritize a seamless connection with Caltrain at Diridon station. 

Reduce Costs, Speed Completion

In total, the tunnel section of Phase 2 represents around 4.5 miles, or 7.2km (see sec 5.3.1). At a cost of $5.6bn this represents a price of $1.3B per mile, substantially higher than the LA Regional Connector, LA Purple Line, or San Francisco’s Central Subway. 

According to this 2018 Citylab article by transit researcher Alon Levy, “Most subway lines cluster in the range of $200 million to $500 million per mile; in Amsterdam, a six-mile subway line cost 3.1 billion Euros, or about $4 billion, after severe cost overruns, delays, and damage to nearby buildings” (see table image below).  Barcelona’s fully automated Line 9 project, from which the single-bore design was taken, was authorized in the same year as the Silicon Valley extension. That project cost ~$300m per mile and opened its entire ~30 mile length in 2016. At that rate, the entire Phase 2 project would come to only $1.3bn, and would have been completed years ago.

For more detail, Levy has conveniently put together a data table with metro costs from around the world.  (Note: the cost estimate shown for the BART SV Phase 2 project is before the recent round of increases).

Given that the original sales tax that was supposed to fund BART to Silicon Valley passed in 2000, it will likely be over 30 years in total from initial funding to completion for the 13 mile extension. By contrast, the original 75 mile BART system was completed only 15 years after the establishment of the BART District.

Institutional improvements needed 

The high costs and extended construction time for BART-SV Phase 2 aren’t just problems with this project.  Similar issues are found in many projects in the Bay Area, where the fragmented nature of our public transportation system, and the need for many individual agencies to do one-off megaprojects, result in expenditures that cost too much, take too long, and may not deliver enough value.

The design choices for this project are affected by fragmented Bay Area governance and divided responsibilities.

  • If BART were spending “its own” money, perhaps $800m in savings would be sufficient to incentivise improved training, but here the construction costs will be borne by VTA, while operational issues are BART’s responsibility.
  • The Santa Clara service and station come from different budgets than Caltrain’s capital and operating budgets. Instead of seeking to reduce redundancy, VTA is seeking to reduce its operating contribution to Caltrain, while proposing a program to completely duplicate Caltrain service on the Peninsula that it would own and control. 

The Bay Area needs an accountable Network Managing agency that plans the network as a network, rather than as individual projects meeting the needs of competing agencies.  

  • A network manager looking to create a coordinated loop of backbone rail around the Bay would not choose to extend BART from Diridon to Santa Clara, duplicating the Caltrain service.  
  • A network manager responsible for the Bay Area’s capital program might consider $800 million in capital savings sufficient to incent additional training, and would consider other ways to reduce costs on this project and deliver more transportation benefits per dollar

Some of the problems with Bay Area megaprojects are exacerbated by the situation where individual agencies build one-off megaprojects, and the region never develops the staff and skill needed to build projects on time and on schedule. A single infrastructure developer for the region’s transit agencies would be a good start, to avoid the agency conflicts documented above. 

But in an era when money will be tight and transit will need to win back ridership, we can’t afford to continue to repeat the mistakes of past Bay Area capital projects that have gone over budget, overtime, and too often haven’t delivered high quality, multi-modal transit facilities that significantly increase transit ridership .