Building a Frequent SMART Line

The currently planned SMART line, while a much-needed addition to our region’s transportation mix, is inadequate as a car replacement. The trains will run every 30 minutes during rush hour, once in the middle of the day, and not at all at night. This is well below the generally accepted 15-minute minimum for show-up-and-go service that you would get on BART. To bring SMART up to that level of service will require an investment, but not as dire an investment as typically thought.

The easiest problem to solve is that of mid-day service. SMART should just run trains during that timeframe, problem solved. Freight could roll during the unused nighttime hours.

The problem of long headways, however, is a physical constraint. SMART operates on a single-track corridor with sidings to allow trains to pass one another as they move in opposite directions. The double-track segments will make up about 17 percent of the corridor, but that’s just enough to allow 30-minute service and not much more.

There are two ways to fix this. SMART could double-track the whole line, or it could increase the number of sidings to match the level of service it wants to have.

Double-tracking: Expensive and possible

To double-track, California law requires a 44-foot right-of-way: 15 feet from the track’s center (centerline) to the edge of the right of way, 14 feet from centerline to centerline and 15 feet on the other side. SMART’s corridor typically includes a mixed-use path as well, which is another 12 feet wide, bringing the preferred right-of-way width to 56 feet.

While most of the right-of-way is wide enough for two tracks and the path, in three locations – Petaluma, Novato, and San Rafael – the width available drops to 50 feet and the mixed-use path will need to be moved to a parallel street. Still, in each of these segments it’s trivial to double-track. In San Rafael, however, we face a different situation. The right-of-way narrows to 30 feet from Puerto Suello Hill to the Downtown San Rafael station, substantially less than required by California for a second track.

Thankfully, the segment is short enough that it doesn’t need one. The 1.8 miles will take about 2.5 minutes to traverse. If we include a 2-minute pad and schedule our northbound and southbound trains to arrive at San Rafael at the same time, there will never be any conflict and therefore no need for a second track.

This solution does introduce some constraints on future SMART operations. Dwell times would need to be introduced to ensure punctuality at San Rafael. Headways could never be less than 7 minutes at current speeds (2.5 minutes for the southbound train to clear plus 2.5 minutes for the southbound train to clear plus 2 minute pad). It might be possible to double-track the tunnel, which doesn’t need as much width, and squeeze out another minute of headway, but by then there would be other problems of capacity that could be solved more cheaply.

The cost-per-mile of double tracks varies from project to project. A double-track project in Carlsbad had a cost of $9.68 million per mile; another project in New York State had a cost of $5.28 million per mile (PDF); and a third in Florida gave about $5 million. These give an estimated cost of between $284 million and $549 million. The lower figure is more in line with industry standards, and it’s roughly half the cost SMART will spend on physical rail on its existing right-of-way.

Sidings: Doing it cheaper

At 15-minute headways, SMART will have at most six trains going in each direction once it reaches full build-out.  If they stick to precise scheduling, they will pass at six predetermined points. Under the current plan, SMART will run 30-minute headways under a similar scheme, with only three passing points of 4 miles each. At that 4-mile standard, we would need another 12 miles of track (another three sidings) to permit 15 minute frequency.  While my original assumption was for 56.7 miles of construction (70.5 miles minus the 1.8 mile Puerto Suello segment minus 12 miles of passing track), with this dramatically reduced need for new tracks we can shrink the cost by a similar margin. Rather than cost $284 million, 12 miles of track will only cost $60 million. Our 7-minute maximum headway will need another 24 miles of sidings on top of that, another $120 million. So for almost half the cost of our full double-tracked system SMART could build the infrastructure needed for exactly the same product.

California regulations treat sidings differently than regular two-track systems, and pegs the minimum width of the right-of-way at 50 feet, rather than 44. While that means the sidings will interfere with the mixed-use path in the narrower segments of the right-of-way, moving the path is far cheaper than extraneous track.

Though this doesn’t give SMART operational flexibility to raise and lower frequencies or speeds at will, the currently planned system doesn’t either. Any changes in frequency or speed will require some capital investment to ensure passing tracks are where they need to be.

The last piece to the puzzle, rolling stock, costs slightly more than the tracks needed. SMART's Nippon-Sharyo DMUs cost $6.67 million per two-car train. At my proposed 15-minute headways, SMART would need 15 trains, 9 more than currently on order, at a cost of $60.03 million. At the maximum service of 7-minute headways, SMART would need 28 more trains than currently on order at a cost of $186.76 million.

The next logical steps – electrification to speed trains, grade separation to eliminate street crossings and automated trains to decrease costs – would squeeze more capacity out of the line, but that’s beyond this exercise.

A high-frequency project is for a Phase 3, not for the current IOS. SMART has yet to prove its worth to the North Bay, and the North Bay has yet to prove it can support a rail line. The density of jobs, residences and activities is currently relatively low near the planned stations. The capital improvements needed are expensive, as are high frequencies, and it’s not clear they would be worth the investment. SMART can’t write off that possibility, however, and needs to engineer its tracks to allow double-tracking in the future. Though it styles itself a commuter rail, SMART could be the primary transit artery for Sonoma and Marin, and it needs to be ready to fill that role if it comes. Until then, the least it could do is run trains whenever it can: 30-minute headways, all day, every day.

A version of this piece appeared on The Greater Marin in two parts: speculating on the possibility of double-tracking, and an update on how to do it cheaper with sidings.