Thanks to a century’s worth of bad regulations, the Bettini Transit Center in downtown San Rafael needs to be rebuilt so SMART can get to the Larkspur Ferry Terminal. The transit center is the busiest transit hub in the North Bay and forms the lynchpin of service in Marin. Getting it right, therefore, is critical, and GGT has four options its considering to do so.Read More
Transit agencies tout themselves as fundamentally “green,” a real solution to global warming and environmental pollution. Every so often, however, an anti-transit activist will complain that buses pollute more than cars or point to SMART’s decision to run diesel trains and say they are just making the pollution worse.As it turns out, the transit agencies are often right, and the anti-transit activists are often wrong, though not always.Read More
When Scott Wiener released his proposal to spur up to 3 million new homes near transit (Senate Bill 827), it received swift condemnation and praise from all corners. Whatever its merits for housing policy – and I happen to think they are quite good – it is not great transit-oriented development (TOD) policy. Good TOD is about far more than housing.Read More
Earlier this year, San Rafael released its report  on parking demand in east San Rafael and found it sorely lacking. There were far too many cars for the space available, leading to overflow into other neighborhoods and constant frustration for its residents. Yet while the report detailed significant outreach and study of the problem, the recommendation for more parking was sadly lacking. Without a discussion of demand management with car sharing schemes or new bike infrastructure, the report could only go so far.Read More
As the historic home of mountain biking the California Wheelmen, and Safe Routes to School, Marin has a unique place in America’s cycling history. Despite that, the number of Marinites biking to work remains quite low and its roads are hardly bike-friendly. What does the research say makes biking more attractive? And how could Marin translate this research into its projects, policies, and priorities?Read More
A common misconception I face is that I don’t like cars. On the contrary, cars are a wonderful thing: they give us freedom of movement, allowing us to go where we want, when we want. They are mobile rooms, so we can keep our stuff relatively safe wherever we go. They are personal mobility tools for people who have trouble walking. They’re a fabulous invention. But they use up too much damn space for everyone to use them.
Or, if you prefer, a diagram of the same, with each box sized according to the space used by a traveller in the mode:
[The data above assumes a Level of Service of D for pedestrians , bicycles , and cars ; 1.2 passengers average for cars; 1 passenger for other private modes; 66-passenger bus; and 2,000 riders on a 10-car subway train.* Bike parking assumes 2 spaces, the size of which were estimated from a 12,500-space bike parking garage in Utrecht ; car and motorcycle parking assumes 3 spaces, the size of which were estimated from planning documents [5a; 5b]; and bus parking assumes a typical maintenance and storage facility . Road space assumes 11-foot-wide urban lanes and 12-foot-wide freeway lanes.]
Jarrett Walker calls is this a problem of geometry: a dense city cannot run on cars . There just isn’t enough space for most people to use it as their primary mode of transportation. Buses, trains, bicycles, and walking all use less space, so they become primary. I’d argue this is not just a problem of dense cities but also of chokepoints within less-dense places.
The first thing I notice is how much parking is needed, especially for driverless and urban cars. It looks like over half the space needs for urban cars come from parking, and almost three-quarters of the driverless urban car needs are from parking.
The goal of transportation planners ought to be to maximize the usefulness of their space. Driverless cars will help cut down on road usage, but their parking needs will still eat up much of the landscape. Driverless taxis would help more, but to be as space-efficient as a half-full bus they would need an average load of 20 people, and at that point it’s more bus than taxi.
In comparison, a bicyclist uses just 7 percent the space of a driver in a city. Someone on foot uses less than 2 percent.
In practical terms, the various space-intensive modes travel should be the most rare ways of getting around, used for practical reasons by people with mobility issues, people who have unusually long or circuitous commutes, and people who have loads to move. Travel by other modes should be much more common, especially at times when a lot of people want to go to a relatively compact area of the region (like at rush hour). These other modes should be just as fast or faster than driving (accomplished by improving those other modes), and relying on them should not feel like a burden.
Far from making driving harder, inverting mode shares so driving becomes more rare would actually make driving easier and better by making other modes much easier and much better. There’s no reason to give away 2,300 square feet of space to every car traveler when 27 square feet by bus or 7 and a half feet by train would do just fine for most trips.
* Trains are the only vehicle here that must operate in their own right-of-way. They can move around 50,000 people per direction per hour using realistic headways compared with 4,800 on buses in their own right-of-way. Without stops, which would approximate the other modes' metrics, a train could operate about twice as often and yield the 7.5 square feet needed per passenger indicated here. A parallel article would examine lane capacity, but that would necessarily exclude parking - a key factor in cars' inefficiency - and so this is not the place for that examination.
 “Current HCM Methodology,” in Pedestrian Level of Service Study, Phase I (New York, NY: New York City DCP, 2006).
 Dan Zhou et al., “Estimating Capacity of Bicycle Path on Urban Roads in Hangzhou, China” (Conference Paper, The 94th Annual Meeting of the Transportation Research Board, Washington, DC, 2015).
 Energy Facility Site Evaluation Council, “Appendix C: Transportation,” in Kittitas Valley Wind Power Project Draft Environmental Impact Statement (Olympia, Washington: Sate of Washington, 2004).
 Juliana Neira, “World’s Largest Bike Parking Garage Opens in The Netherlands,” Designboom, August 10, 2017.
[5a] “Design and Improvement of Parking.,” Section 19.26.040, Roseville Municipal Code, accessed August 18, 2017; [5b] Peter Croft, “Light Vehicle Sizes and Dimensions: Street Survey Results and Parking Space Requirements – Information” (Auckland, New Zealand: Land Transport NZ, December 2004).
 Division of School Support, “School Bus Maintenance Facility Planner” (Raleigh, NC: North Carolina Department of Public Instruction, February 2011).
 Jarrett Walker, “Does Elon Musk Understand Urban Geometry?,” Human Transit, July 21, 2016.
The dream of urbanites across the central Bay Area, as they gaze out towards the golden hills of the Marin Headlands, is BART. They share that dream with the suburbanites who stare back at the picturesque view of San Francisco. That it didn’t happen is now simply a fact of life in the Bay, but we need not live with this fact.
Two weeks ago, I tackled the science of traffic congestion: why it happens and the damage it does to our transportation system . Last week, I examined the best way to cope with congestion – an anti-congestion toll – and how to craft anti-congestion policies with an eye toward equity in Marin .
Yet an anti-congestion tolling plan works best when there are effective alternatives. Golden Gate Bus is, when it doesn’t face traffic, a fast and efficient mode of transportation. Unfortunately, it doesn’t run often and it runs into traffic frequently, especially within The City. As well, transfers between it and other long-distance transportation systems like BART and Caltrain are poor at best. Marin and Sonoma residents, then, don’t have viable alternatives to driving if their destination isn’t downtown San Francisco.
The plan in a nutshell
SMART South, as I call this plan to differentiate it from the SMART that is currently under construction, consists of the completion of the SMART system, the upgrades needed to operate at eight trains per hour per direction, and the upgrades, construction, and new trains needed to run SMART south to San Francisco and as a subway under Geary Boulevard.
This will include electrification of the existing line ($125 million  for 39 miles of track), pushing SMART north to Healdsburg and south to Marin City ($537 million [i] for 40 miles of surface track), rebuilding the Alto Tunnel ($60 million ), tunneling from Marin City to the Golden Gate Bridge ($850 million [ii] for 1.7 miles of tunnel), retrofitting the bridge ($392 million ), tunneling from there to the Geary Boulevard Subway ($1,365 million [ii] for 2.7 miles; the Geary Subway segment would be built separately from this project) adding passing track to the existing and new surface line to allow for higher frequency ($705 million [i] for 52 miles), elevating the downtown San Rafael track ($93 million [ii] for 0.6 miles of track), finishing the bike path from Marin City to Cloverdale ($5 million [i]), adding the Corte Madera/Larkspur, Mill Valley, Marin City/Sausalito, Vista Point, and Presidio stations ($500 million [ii]), and buying Caltrain-compatible bi-level electric trains to allow for a much and more frequent longer line ($654 million  for 38 3-car trains). There are also $250 million in miscellaneous costs associated with finishing the SMART line as promised. In total, this comes to about $5.5 billion for an effectively-new 85-mile system or about $65 million per mile. That’s quite a steal for American construction costs.
Passengers would be able to transfer at Union Square to BART and Muni’s under-construction Central Subway, and at the Transbay Terminal to Caltrain and high-speed rail. Depending on how the second transbay tube is built, passengers would be able to move on either to downtown Oakland or south to the Oakland Airport and Fremont without transferring.
The short of it is that anti-congestion charges on the Richmond-San Rafael Bridge (RSRB), Golden Gate Bridge (GGB) and the Novato Narrows could be leveraged into $1.4 billion [iii]. A 0.4% sales tax in Sonoma, Marin, and San Francisco could be leveraged into another $1.3 billion [iv], for a total of $2.7 billion raised locally. Regional, state, and federal monies cover about half the cost of major transit construction projects in California, and that’s the balance of the project. Ongoing operations & maintenance would be covered by fares, tolls, and sales tax income as well as state and federal support. The gritty details of all this, including a speculative financing plan and how to calculate anti-congestion tolls, will be in an upcoming post for the people who are truly interested.
SMART South as envisioned would be built alongside a secondary transbay project extending Caltrain and some kind of heavy rail subway across the bay to Alameda County. These would happen regardless of SMART South, so its value is really in connecting the North Bay with the South and East. How many riders would use that link?
Last year, I examined what higher frequencies could do for the base SMART system’s ridership and arrived at a conservative 12,100 daily trips. Adding the southern Marin, downtown Novato, and Presidio stations would push the ridership up to 22,000 using the same model .
Using 2014 Census data, it appears that roughly 7,500 people live within a half-mile of SMART and SMART South stations and work within a half-mile of BART or Caltrain, or vice-versa . While only half of them are likely to commute via transit after this extension, additional riders would come from outside the half-mile radius, and commutes only account for about 20 percent of all trips. Wrap all that up and we have another 50,000-80,000 trips per day, depending on fares, travel time, and transfers. Given that this range comes from a model that assumes no anti-congestion tolls - and that such tolls boost transit use – I’d lean more towards the higher than the lower number. That would nearly double the capacity of the Golden Gate and northern Highway 101 corridor.
So for about $5.5 billion, paid for with new tolls and a 0.4% sales tax, SMART could dip south into San Francisco, fully tying Sonoma and Marin into the regional and statewide rail systems. It would provide a viable alternative to the freeway. Alongside tolling, it would make congestion a thing of the past in the North Bay, making deliveries more timely and commutes much more reliable.
This is merely a draft, of course, and should be subjected to more rigorous study. The political difficulties of yet another sales tax and yet-higher tolls are apparent. But SMART South is eminently attainable. With a little ingenuity and a little optimism about the future, it could be our next stop.
[i] Estimate from existing SMART costs.
[ii] Estimate from a variety of peer projects.
[iii] Assuming borrowing is on the same terms as MTC’s toll-related bonds and is spaced according to construction needs.
[iv] Assuming borrowing is on the same terms as SMART’s sales-tax backed bonds.
 David Edmondson, “The Science of Traffic and Its Awfulness,” The Greater Marin, July 3, 2017.
 David Edmondson, “Let’s Get into the Weeds: A Congestion Charge Plan for Marin,” The Greater Marin, July 10, 2017.
 Stephen A. Gazillo, “A Planner’s Guide to Fixed Guideway Electrification Projects,” Transportation Planning, November 2005.
 County of Marin, “Investigative Study to Begin on Alto Tunnel” (County of Marin, January 10, 2017).
 Charles Seim, Mark Ketchum, and T.Y. Lin International, “Golden Gate Bridge Mass Transit Feasibility Study” (San Francisco, CA: Golden Gate Bridge, Highway and Transportation District, October 1990).
 Railway Gazette, “Caltrain Signs Double-Deck EMU and Electrification Contracts,” Railway Gazette, August 16, 2016.
 Transportation Research Board, “Elements Needed to Create High Ridership Transit Systems,” Transit Cooperative Research Program (Washington, DC: Federal Transit Administration, March 2007).
 Center for Economic Studies, “Longitudinal Employer-Household Dynamics” (Washington, DC: United States Census, 2014).
Last week, we discussed how tolling designed to eliminate congestion would improve travel times and improve the efficiency of our roads.  This week, we get a bit into the weeds about how such a plan would work.
How to eliminate congestion in the North Bay
Marin has four main entrances, three of which have significant rush-hour congestion: the Novato Narrows, Highway 37, the Richmond-San Rafael Bridge (RSRB), and the Golden Gate Bridge (GGB), with Highway 37 being the odd one out of the congestion game.* These congested gateways are a drain on people’s time and public resources. The Narrows is only moving 85 percent of the vehicles per hour that it could, the RSRB moves just 61 percent, and the GGB moves just 74 percent capacity.**
The entrances are all chokepoints, with few or no parallel routes, so they are ideal for tolling. The toll need to be charged both directions, so existing tolls on the GGB and RSRB would be cut in half: half heading inbound and half outbound.
A toll that eliminates congestion would allow another 4,380 vehicles per hour to move at rush hour, a staggering 35 percent increase over today's numbers.
To determine the exact amount a congestion toll ought to be, we need some pieces of information: the maximum number of vehicles per hour each road can move, also known as their flow rate; the existing flow at peak hour; the amount of demand there is at peak hours; how long the flow is congested; the length of the congestion; and the median income of the area. If all of this is known, then a little calculus is needed and then, voilà, the result. Unfortunately, with publicly available data it is very difficult to determine demand, so we need to work a little bit harder for a less precise number.
I wanted to find out how much extra time people were spending in traffic than they would if all the roads were at their maximum flow and how much that lost time is worth. Tie these elements together and we get some estimates. The Narrows needs a charge of about $5 ($108 extra per month), the RSRB needs $5 on top of its existing toll (, and the GGB needs $6 on top of its existing toll. Combined, these tolls would bring in an additional $102.4 million per year, which could be used to secure bonds of up to $1.3 billion (over 30 years at a 4 percent interest rate). For context, the Golden Gate Bridge brought in $129.5 million in toll revenue in the 2015/2016 fiscal year. 
Because Marin’s transit system is part of a broader regional system of bridges and highways, any tolling plan would need to be regional in order to ensure that traffic doesn’t spill over onto some other road. As previously mentioned, Marin’s entrances have very few good roads and therefore can be fairly easily tolled. However, if Marin manages to unclog its traffic, East Bay commuters might try to cut through the county on their way to San Francisco. More drivers on Marin’s roads means higher tolls, and then Marinites and Sonomans would have to pay more than before. Any toll plan would need to be implemented across the whole region to ensure all the region’s freeways are used as efficiently as possible and nobody is charged too much.
Build equity into the toll
One of the oft-cited problems with charges like this is that it is regressive as it hurts the poor, who have more time than money, more than the rich, who have more money than time. I outlined a solution briefly last time,  but here it is worth fleshing out a bit more.
Carbon taxes are often structured in such a way so as to be revenue neutral,  so whatever income is generated is refunded to the users. However, an unclog-the-roads toll, like what’s proposed here, should be used to increase the tolled travel corridor’s capacity and so needs to provide a useful income. This could be done either by selectively exempting vehicles registered to poor drivers, but that would be logistically difficult. Instead, tolls should be a refundable state tax credit that phases out as income increases. (Refundable tax credits are refunded to the taxpayer whether or not she owes any money, and so they are always returned.) FasTrak, or whoever the vendor is, would inform their registered users every year how much they paid in congestion tolls, which the users could report on their taxes. Non-registered users, like tourists or infrequent users, would not get a form so they’d end up paying no matter their income.
It’s difficult to estimate how much of the toll income would be refunded, but given that drivers tend to be wealthier in general, especially in the suburbs, this may not be a significant portion of toll revenue. Should it become viable to issue monthly rather than annual refunds to poor drivers, it would likely encourage driving, which would in turn increase congestion. That would require higher tolls, offsetting the revenue lost to refunds.
Regardless, because the toll essentially is trading time for money, and the amount of time currently used up in traffic is worth about $102 million per year, that is roughly how much would come in from these tolls.
Next time, we will consider just what might be done with this revenue stream to make travelers lives better.
* Traffic from 37 does add to congestion on Highway 101 to the south, and congestion does regularly crop up on the road, but the MTC analysis this post relies on  reports very little congestion on the road itself. Without a proper traffic management analysis, it would be tough to say how these tolls would actually shift congestion around the North Bay.
 David Edmondson, “The Science of Traffic and Its Awfulness,” The Greater Marin, July 3, 2017.
 Golden Gate Bridge, Highway, and Transportation District, “Comprehensive Annual Financial Report,” Annual Report (San Francisco, CA: Golden Gate Bridge, Highway and Transportation District, 2016).
 Edmondson, “The Science of Traffic and Its Awfulness.”
 UNFCCC, “Revenue-Neutral Carbon Tax,” Momentum for Change, accessed July 4, 2017.
 Metropolitan Transportation Commission, “Time Spent in Congestion,” Vital Signs, accessed April 29, 2017.
 John D. Zegeer et al., “Default Values for Highway Capacity and Level of Service Analyses,” National Cooperative Highway Research Program (Washington, DC: Transportation Research Board, 2008).
 Division of Traffic Operations, “2015 Traffic Volumes on California State Highways” (Sacramento, CA: Caltrans, 2015).
If only the guy in front of you would drive better, perhaps you wouldn’t be stuck in traffic. Slow, fast, break. Slow, fast, break. Shift lanes. Whoever it is, that driver sucks.
That’s what the driver behind you thinks about you, and what the driver behind that driver thinks about him, and so on. It’s traffic congestion, and it’s bad for everyone.
There are two aspects of congestion to be explored here in very brief detail: its physical results and its economic causes. Only once we understand the economic causes will we be able to figure out a solution.
Traffic makes roads terrible
Roads move traffic at a rate governed by a very simple model: the density of the vehicles per mile times their speed. But there is a trade-off. As the density of vehicles goes up, their speed goes down as drivers become more cautious. There’s a balancing point where density and speed allow a road to move the most vehicles. Shift the density a bit either way, and you don’t move as many vehicles, and you’re not getting your money’s worth out of the road. The relationship looks like this:
So, traffic congestion is when the road is operating somewhere on the right side of the curve. Basically, at rush hour, there are too many people that want to use the road all at once, increasing density way past the efficient level, making everyone slow down and the roads awful. This video does a good job of showing how this works. It’s boring, but stick with it :
Basically, everyone slows everyone else down by trying to use the road all at the same time. Every additional vehicle on the road slows everyone else down just a little bit, which adds up to a lot. Past the balancing point, each additional vehicle also makes the road less efficient. Just how much is something we can quantify.
Time is money
When someone is slowed down in traffic, they get to wherever they’re going later than they’d hoped. This might cost maybe a few minutes, but when multiplied out over everyone stuck in (and causing) that traffic it ends up being a lot of person-hours. Economists, measuring a whole host of factors , translate this time cost into a monetary cost. Every vehicle “pays” a certain amount of time (alongside gas, maintenance, and tolls) which is the average cost of using the road. When someone new comes into the road, that average cost goes up just a little bit at the margins. This is called marginal cost of using the road. Each driver feels only the average cost; the marginal cost is divided up evenly between everyone else around them on the road. It’s like when someone cuts in line: everyone behind them is slowed down a bit.
In short, every vehicle faces the average cost and forces everyone else to “pay” a little bit more. Brendan O’Flaherty writes:
The final major kind of external cost that drivers impose on each other is congestion. On an otherwise deserted country road, or even on most city streets at four o'clock in the morning, it's difficult for one car to get in the way of another car or to impede its progress, and so congestion is not a problem. But when tens of thousands of cars an hour are converging on the Lincoln Tunnel during rush hour, they get in each others' way, and since the number of cars that can go through the tunnel in any minute is less than the number that want to go through, each car that enters the tunnel during rush hour is forcing all the cars behind it in the queue to wait a few seconds. Because drivers don't bear the cost of the congestion they cause, they cause too much of it.
I'm not saying that drivers don't bear the costs of congestion. Of course they do; being stuck in a traffic jam is unpleasant and time-consuming. But I don't bear the costs of the congestion I cause; instead I bear the costs of the congestion you cause, and so you have no incentive not to cause it. Or more starkly, think about a line of cars going through the Lincoln Tunnel. The first car may encounter no congestion, but if it were vaporized, every other car would get through the tunnel more quickly; and so the first car is causing a lot of congestion. The last car, by contrast, is encountering a lot of congestion, but causing none (if it were vaporized no one else would get through the tunnel sooner). 
O’Flaherty goes on to also describe how people shift their schedule around to avoid traffic, which has its own cost, and how people sometimes just grin and bear it, choosing to drive in traffic and pay that cost on the road instead. 
But all this cost is just in terms of time. It’s wasted, lost to everyone. If there were a way to shift that cost into dollars, then it would be possible to put that cost to productive use, or to return it to the people who need money more than time.
This is where congestion pricing, or demand pricing, or surge pricing, or whatever you want to call it, comes in. In essence, it’s a toll that charges people what they would otherwise pay in terms of time as well as the cost they impose on other users of the road (the marginal cost). This smooths out demand, so the road never gets clogged, and allows the money to be put to better use. The tolls collected could also be returned to poor drivers as a refundable tax credit. (I would have the electronic toll collector contractor, i.e. FasTrak, mail out a tax form to each of its customers itemizing the amount paid in anti-congestion tolls by month, which the customer could report on their taxes as a credit contingent upon income.)
The goal would not be to return the road to free-flow speed, but rather to the balancing point, where a road moves the most cars per hour possible. More people would therefore be able to get through the system at rush hour, meaning the whole driving system would be more efficient. Any money not returned as a tax credit could sponsor public transit or road improvements.
The London tolling plan, which did not include the tax credit, saw travel times fall and then stabilize within the controlled area, though traffic volumes continued to fall significantly.  London also saw the number of crashes and traffic deaths fall significantly.  The Stockholm plan saw the similar results but, intriguingly, drivers who were interviewed after the toll went into effect didn’t think they had changed their travel behavior. 
This is the only way to ensure congestion doesn’t occur without a recession because, as discussed above, people normally don’t feel the cost they impose on everyone else on the road and so they overuse it. With a toll like this, people will feel the full cost of their road use and so allow society to get the most out of the investment in roads.
Of course, people don’t like to be charged for what they previously got for free, especially when that would represent an unknown change. Stockholm residents, for instance, hated the idea of a congestion charge until it was attempted.  Manhattan has tried for years to implement a charge only to be blocked by state lawmakers.  When San Francisco talked about doing something on the Golden Gate Bridge, then-San Rafael mayor Al Boro called it a “Marin commuter tax.” 
Nevertheless, given that it is the only way to permanently resolve traffic, it is worth exploring how big the charge would be and what is possible with that income. That’s what we will explore next time.
 J. Hewitt, “The Calculation of Congestion Taxes on Roads,” Economica 31, no. 121 (1964): 72–81, doi:10.2307/2550927; Roberto Ayala, “The Value of Travel Time Savings: Departmental Guidance for Conducting Economic Evaluations Revision 2 (2014 Update)” (Washington, DC: US Department of Transportation, July 9, 2014).
 Transport for London, “Public and Stakeholder Consultation on a Variation Order to Modify the Congestion Charging Scheme Impact Assessment” (London, UK, January 2014).
 Alex Davies, “London’s Congestion Pricing Plan Is Saving Lives,” Website Type, Wired, (March 10, 2015).
 Ben Fried, “Factchecking Cuomo’s Revisionist History of NYC Road Pricing,” Streetsblog New York City, February 18, 2015.
Update: SMART just reached out to me with news that they are looking at the schedule again based upon the feedback they've received, so stay tuned for a new schedule.
By now, word of the final SMART schedule is out, and the response is relatively negative. The schedule (PDF here) has a 90-minute gap right at the heart of the rush hour, between 7:26am and 8:56am. Folks were not having it on Twitter. These two tweets are fairly representative:
The SMART Board wasn’t happy with the results either. San Rafael Mayor Gary Phillips told staff at the meeting where the schedule was released, “With these kind of gaps I’m concerned people will say, ‘I’ll continue to drive my car,’ I would encourage we revisit this” (1).
For their part, SMART spokeswoman Jeanne Belding wrote me indicating the schedule was a result of a survey of potential riders after a draft schedule was released. She added, “The schedule we are opening with is based on previous feedback, our staffing, and takes into account the fact that we have a single-track system” (2).
SMART General Manager Farhad Mansourian indicated that the reason for the schedule is, in short, that it is complicated: with the system running as it is, a change to one train means all trains must be changed. And there’s freight trains to deal with, only four active passenger trains, and so on (1). But I think there is more to discover in the schedule, and that the system is not nearly as hamstrung as GM Mansourian seems to indicate.
Let it be known that just because the schedule is bad doesn't mean SMART isn't operating under real constraints. Undoubtedly, this was someone's attempt to polish the poop (or shine the shit) that is their situation. Nevertheless, this is definitely not sufficiently shiny yet, and I think we can do better. Let's look under the hood at what we're dealing with.
First, we can divide the schedule into a series of what I call platoons: four trains, each departing 30 minutes after one another, followed by a minimum of a 90-minute break, followed by another platoon. These are set by the number of trains SMART can run at any given time: just four of their seven-train fleet. Spokeswoman Belding told me they have had trouble hiring drivers, citing the high cost of living as the reason, and so they are limited to just four trains on the tracks at any given time. She did not answer a question about when they expected to be at full strength (2).
Another limiting factor are the passing tracks. Taking best practices from Switzerland and Germany, SMART built a single-track commuter system with four four-mile-long passing tracks at key locations so trains could run in both directions and pass one another. The problem, of course, is that these passing tracks limit the schedule to certain ranges. What those ranges are, SMART can’t say, but we can assume that the schedule as presented is one solution.
Taking all this into account, we can trust that changing one train’s arrival or departure time more than a few minutes requires the rest of the schedule to be changed by the same amount.
So now, the problem at hand: what to do about that 90-minute morning gap.
I took this as gospel and rotated the entire schedule around the arrival time of Train 5 at downtown San Rafael. Here’s what I call San Rafael Rush:
The San Rafael Rush schedule gets everyone to downtown San Rafael before 9am and ensures people still have time to walk to work after the last train arrives. Unfortunately, it makes things tough on northbound commuters, giving very little flexibility for Marinites working in Sonoma.
The second option is what I call Early Bird:
This option pushes everything back by about 10 minutes, ensuring that the 90-minute gap ends right as rush hour is at its worst. I’m not so keen on Early Bird, as few people arrive at work during those early morning hours served by the first platoon, but it does make sure that people can at least get to the office by 9am.
There are other methods of making the schedule fit, but it’s important to understand that SMART is operating in a tension between what riders say they want versus what the data says they should do. Census data shows what we know: that a plurality of people working in San Rafael arrive between 8am and 9am, people whom the current schedule poorly serves. SMART also insists that its schedule is based upon feedback from its own scheduling poll of potential riders.
It’s also important to note that GM Mansourian’s statements that nothing can be done are obfuscation. True, SMART is limited in its schedule movement, and they will be until they get the drivers they need, but they can indeed do better than what they’ve given. There is a bit of hope on that front. The GM indicated they’d be open to changing the schedule based on feedback after the train has started running. “This is what we will test and get feedback. If we can tweak, we will tweak” (1).
PS - And just because you've heard it a thousand times already, I did promise to remind you not to risk getting hit by a train. Sure, they'll be far less dangerous than drivers, but that doesn't mean you want to be the guy making everyone late for work because you got squashed by a train. Also, I will take this opportunity to repost one of my favorite ads, Dumb Ways to Die.
(1) Prado M. SMART schedule has major service gaps. Marin Independent Journal [Internet]. 2017 May 17 [cited 2017 May 22]; Available from: http://www.marinij.com/article/NO/20170517/NEWS/170519805
(2) Belding J. Service schedule & trains. 2017.
Public transit is not just for people in big cities. All across the United States, public transit is a lifeline for millions of people who cannot afford a car or cannot drive. Unfortunately, these routes often provide the barest minimum of service: four round trips per day, no Sunday service, early end times. To add insult to injury, they are very rarely mapped in a comprehensible way, forcing riders to study dense timetables and obtuse maps.
To solve this problem in Tompkins County, New York, I created a frequency map indicating all stops, all routes, and major connection points.
The existing maps are quite poor: there is no system-wide map, forcing riders to flip between pages to follow a single line along its course. Service levels aren't indicated, with weekend or late-night service shown as being just as important as weekday or high-frequency service. Often, new riders will avoid buses they don't know, even if the bus is going where they want to go, simply because there's no way to know where it will take them. Further, the local system, TCAT, has flag-stop service, where riders can get on or off a bus wherever the driver can safely pull off the road, but it is not indicated. In talking with riders, my team and I found people often didn't even know about the service. The existing maps' final sin is that it groups routes together on the same line even when those routes have nothing in common.
The new map fixes each of these problems. It adjusts line thickness to indicate mid-day frequency; groups late-night and weekend routes together with their weekday counterpart routes (the 70 is the weekend counterpart of the 30, so these are combined into a single line); indicates flag-stop zones; and is envisioned as showing the full county's system in a single panel. For rural riders, for whom roads are critically important to wayfinding, the road grid is underlaid behind the entirety of the map.
I started a similar project a few years ago for Marin and Sonoma, but my design skills were simply too poor to finish. It taught me a lot, but, given the major changes since starting this project, I suspect I would need to start again. More interesting would be a better map of Bay Area transit service:
Bad design is not necessarily the fault of the transit agency - good designers are expensive and need management, and often staff and budgets are already too small. There is little time or energy to spearhead change. Still, they ought to consider the benefits that come from comprehensibly conveying the entirety of the system in one clear image.
In this week’s IJ Forum, guest and transit skeptic Richard Hall brought up something I didn’t expect: the argument that transit frequency equals freedom . This is something that transit advocates have been harping on in Marin and elsewhere for years, and one of the key factors that is likely to hold back SMART from reaching its potential. Though I addressed this point back in 2012, with the train opening up this year it is worth revisiting how SMART might be able to reach higher frequencies, and how much it might cost.
There is blood in the single-track stone
Once operational, SMART’s trains will run in both directions simultaneously despite having only one track. It can accomplish this through the judicious use of passing tracks, meaning a strategic stretch of the corridor will be double-tracked so trains can pass one another.
At full build-out, from Cloverdale to Larkspur Landing, there will be 4 such passing tracks, each of which is 12 miles long, which will allow trains to run every 30 minutes in either direction. (This 30-minute time is called the “headway” in transit-speak.) To double the frequency to 15 minutes, we would need to double the number of passing tracks to 8. To double frequency again to 7.5 minutes, we would need to double the number of passing tracks again to 16.
At this point, the permanently single-tracked Puerto Suello Tunnel between North and Central San Rafael becomes the choke point. Any further increases would need to widen the Puerto Suello Tunnel and basically double-track the whole system. This would likely overkill for the foreseeable future.
Using the existing construction costs as a guide, the cost for the passing tracks on a 15-minute headway system is $60 million. The cost for a 7.5-minute headway system is twice that: $120 million . If we include the cost of the trains needed to run such frequent service, the total cost would be $120 million and $307 million, respectively .
We can squeeze blood out of this stone if we are willing to spend the money for it.
Of course, ridership at this point becomes a problem. Is it worth it to run trains this often? Definitely.
There are almost 43,000 jobs and over 19,000 people within a half-mile of SMART’s stations. 2,237 people live and work within the station areas , but commute trips account for only about 20 percent of all trips, and this doesn’t include people who might use the park and ride services. Ballpark figures, using a ridership model [PDF] from the Transportation Research Board put the ceiling of SMART’s potential ridership at about 25,000 . SMART itself estimates its ridership will be around 4,500 trips per day, which the model agrees with.
We need to dive a little bit into some wonky economics for a moment using these numbers but bear with me.
When a transit operator alters some aspect of a transit service, they of course also alter the ridership. Increase the vehicle speed and ridership goes up. Increase fares and ridership goes down. This is called the elasticity of demand: how much does a given change affect the demand for that transit service. The first example mentioned references the in-vehicle time elasticity because it indicates how much people change their habits based on how much time they spend in a vehicle. The second example is the price elasticity because it deals with the price of using the service.
Headways alter the access time elasticity, which is more valuable to people when it comes to transit. As much as you may enjoy your walk to the train, you probably enjoy the fact that you can work or sleep on the train en route a little bit more.
Access time elasticity among commuters is 1.28, which is pretty high. Decreasing average wait times by 1 percent by decreasing headways by 2 percent increases ridership by 128 percent . Doubling SMART’s frequency to 15 minutes, then, will get us a 64 percent* increase in passengers: from 4,500 trips per day to about 7,380. Doubling frequency again could get us to 12,100 trips per day. This, of course, does not count the number of people who would use SMART for non-commute trips too, and it also ignores the effects of improving on the very infrequent service SMART is planning on providing in the middle of the day.
I would argue that a 15-minute frequency would absolutely be worthwhile on the basis of the commuters alone (the cost per new commute trip is $42,000, half the cost per trip on the baseline SMART system). A 7.5-minute system is not as cost-effective based on new commuters, but is significant if we include non-commute trips.
It may also be a viable alternative to a wider Novato Narrows. The traffic congestion there is in part due to a 15 percent increase in the number of cars travelling through to Central San Rafael, or roughly 500 more vehicles per hour . Diverting 3,690 trips per rush hour (half of 7,380) would alleviate that congestion, at least until drivers fill up the space again thanks to induced demand .
So Hall is quite right on this point: transit frequency influences how people travel and how many people use the system. I am unsure whether he knew just how influential frequency can be, but no matter. SMART would do well to examine the effects of increasing its service frequency.
* Average wait times are half of the headway: if you arrive at a random point during the inter-train period, your average wait is half the full headway. Doubling frequency, then – a 100 percent increase – decreases average wait times by half that, or 50 percent.
Note: Given how off-topic any comments section can get in a SMART-focused article, I have turned them off for this post.
 Traffic in Marin, IJ Forums (San Rafael, CA, 2016).
 David Edmondson, “High SMART Frequency on the Cheap,” The Greater Marin, August 8, 2012.
 David Edmondson, “Can SMART Double-Track?,” The Greater Marin, August 6, 2012.
 Center for Economic Studies, “Longitudinal Employer-Household Dynamics” (United States Census, n.d.).
 Daniel G. Chatman et al., “Making Effective Fixed Guideway Transit Investments: Indicators of Success” (Washington, DC: Transportation Research Board, 2014).
 Arthur O’Sullivan, Urban Economics, 8th ed (New York, NY: McGraw-Hill/Irwin, 2012), 295.
 David Edmondson, “The 101 Corridor: Transportation Myopia in Practice,” The Greater Marin, January 13, 2013; Caltrans, “Traffic Volumes on California State Highways” (Sacramento, CA: Government of California, 2014).
 Connor Jones, “The Street Economics of Induced Demand,” The Greater Marin, December 21, 2015.
Shuttle service to SMART’s temporary north end will be winding, slow, and inefficient. That’s bad news for Sonoma’s towns, which are already under strain from car commuting.
Sonoma County is getting passenger rail service for the first time in decades, hopefully starting at the tail end of 2016, with the opening of the SMART train.
SMART is starting to move from being an agency building a train, to an agency that will *run* a train. Big difference. Based on a lot of things I have heard in the past, there is concern that General Manager Farhad Mansourian is a better project manager CEO than an operations CEO. Time will tell, but as we approach the opening and decisions start to fall in place, I am going to reload this blog and follow the topic.
At last month's SMART board meeting, there was a presentation on First/Last Mile connectivity. The PDF is linked above, there is also avideo of this meeting - March 2, 2016 which is interesting and a bit illuminating. The board rightly gives the staff of SMART and the related agencies credit for a lot of hard work, but there are a lot of holes in the strategy which underscores that transit agencies and boards don't really focus test their ideas. They look at a problem, place themselves in the problem, and imagine how the problem needs to be solved. Witness Caltrain discussing workers who can go in "later" because they have "flexible" schedules. They go to work 8-5, my office doesn't even turn the lights on until 9 AM and people get upset at meetings before 10. The schedule isn't "flexible", it's flat out different.
With SMART, the board (mostly local politicians) and staff are working from a very "how do I get to San Francisco" mindset. I can't blame them, a lot of the public reacts this way as well, if you read internet comment boards. Officially, SMART diverges from this message on their website:
Today, more than 75% of commuters in the North Bay travel either within or between the two counties to get to work.
Thus we get to my first topic on the presentation on first/last mile - the North County Coordination to be provided by Sonoma County Transit. This connection is at the direction of the board and not negotiable - Windsor, Healdsburg, and Cloverdale were on the original proposed train line, and have had train service delayed indefinitely due to lack of funding. There is a bus from Cloverdale to Santa Rosa - route 60 - but it's slow compared to the freeway and doesn't go to the train depot on Airport Rd. SCT is going to add a shuttle - as seen in the first/last mile presentation, that will express between the 3 towns and the North SMART terminus at Airport Road. It will mostly run on US-101 in uncongested areas making it a quick connector. Sort of.
The proposed schedule shows the "go south in the AM, north in the PM" mindset of SMART. The only shuttles in the AM run North to South to meet trains, there are no proposed shuttles the opposite direction. One problem - the City of Healdsburg is a net *importer* of labor, not an exporter! SMART is trying to figure out how to get a small population of Healdsburg residents south, instead of the large population that is trying to get TO Healdsburg. The population shift during the day is such that tiny Healdsburg is starting to have parking wars and is now considering using valuable downtown land to build more parking.
Housing prices in Healdsburg are forcing the town's workforce to leave the city for Santa Rosa and Cloverdale, less expensive areas, which is bad on its own, but exacerbates the parking problem which leads to bad land use decisions which feedback to make the housing problem worse. That workforce could theoretically take SMART to the Airport and hop a shuttle to town, but it won't exist.
Not only does a northbound shuttle not exist, but the market of workers in Cloverdale who could take the shuttle to Healdsburg will be poorly served. The express shuttle will go from Cloverdale to Healdsburg in 20 minutes. SCT route 60 takes ~40 minutes to make the same run - the express cuts the trip in half and could attract new riders. However, because SMART and SCT are only thinking about "get people to the train" - the stop is located at the decrepit Healdsburg train depot on the outskirts of town (and they are building a $1 Million parking lot there), producing a walk for people making that trip which eats up any time savings. It also means that any tourists from SF who decide to take this route get dumped off in the middle of nowhere instead of the middle of town. Might as well drive. This in order to provide park and ride service to a bus for Healdsburg residents? The Healdsburg depot is out of the way for most Healdsburg residents, the bus will make an additional out of the way stop at the Windsor "Train Depot", before winding to Airport Road. Summary - any sane Healdsburger with a car will simply drive to the Airport Road Station.
The detour through HBG to get to the old depot, and the similar winding trip in Windsor kills the trip time, reducing any incentive for people coming from Cloverdale to use the shuttle. The buses should make quick stops just off the freeway but close to the downtowns - the Amtrak bus stop at Mill/HBG Ave in Healdsburg, and right off the freeway in Windsor next to where there is a McDonalds. Shuttle service like this relies on speed. Optimizing it is the only chance to get the ridership needed to keep the shuttle going and hopefully support the train. This includes understanding that some of the riders will eschew the train altogether, using the shuttle as a fast intra-North County bus service. And they should provide service in both directions at both peaks.
Original Post: Murphy, John. 2016. “SMART Train - Last Mile Connectivity - North County.” Holier than You Blog. March 31.
Other Works Cited
Healdsburg, CA. 2016. “Cerri Site (Purity Building) Redevelopment Project.” Healdsburg, California Official Site. Accessed May 29.
Mason, Clark. 2015. “Mass Evictions in Healdsburg Prompt Cries of Racism.” The Press Democrat, July 8.
Michel, Franck. 2014. Waiting for the Bus. Photograph.
Sonoma-Marin Area Rail Transit. 2016a. Board of Directors Meeting: March 4, 2016. Petaluma, CA.
———. 2016b. “First/Last Mile Connection.” Petaluma, CA, March 2.
U.S. Census Bureau. 2016. LODES Data. Longitudinal-Employer Household Dynamics Program.
Despite years of arguing new housing will exacerbate traffic congestion and blaming the county’s urbanists for indifference to the problem of congestion, Marin Against Density (MAD) is now fighting against new traffic lanes on Sir Francis Drake (Drake), arguing the project will actually reduce capacity. Fact check: it won’t.
Drake Boulevard is a mess south of Ross from every perspective. To people on foot or bike, it’s a hot, loud, dangerous traffic sewer. To people in cars, it’s a congested nightmare every rush hour. Thankfully, there’s enough space on the road to make the first problem a little less bad and fix the second problem.*
The biggest change to Drake would be a third lane heading south from Marin Catholic to 101 and, in one design, north from 101 to El Portal by narrowing lanes from between 15 and 21 feet to a standard 11 feet. This will increase capacity by about 50 percent along the most congestion stretches of the road, a huge boost for drivers that have complained about for years.
Other changes would be squaring off intersections in a few sections to make it safer to walk, new shoulders – formerly Class II bike lanes – and wider sidewalks . Given that there are schools along the route, these are victories for parents and children along the way.
So what’s the problem?
According to an email from MAD , and to comments from anti-urbanists from the last time I tackled this problem in November , the primary complaint is that narrowing lanes will reduce capacity by slowing traffic. They’ve labelled it a traffic calming exercise, designed to support safety rather than traffic flow.
Even if we look past the morally reprehensible attitude that the safety of people walking isn’t worth protecting, it’s utterly illogical. MAD doesn’t understand how roads work.
A short primer on road capacity functions is in the notes, if you’re interested, but here’s the short of it: moving from a 12-foot lane to an 11-foot lane will decrease the free-flow speed of a road by about 3 percent, but adding a whole new lane will increase capacity much more than what’s lost by that very slight depression in free-flow speed. Add it all up, and the project should boost capacity by 45.5 percent.
But what about right-turn lanes lost? There is still more road capacity, so while the far right lanes might be slower during off-peak times, through drivers who won’t be in the far right lanes won’t need to worry.
To put it very simply, the county wants to slow uncongested travelers by about 3 percent to add 50 percent capacity. This seems like the kind of project MAD would support if it was so worried about traffic.
But MAD’s opposition to adding makes so little sense it boggles the mind. Why would an organization that argued any new homes will cause congestion fight against measures that might actually reduce congestion? The political answer might be the easiest.
It’s an election season, and that Fairfax email was rife with disparaging words for incumbent supervisor Katie Rice and glowing words for the conservative, Kevin Haroff, who has come out against the project . By painting this redesign as a road diet rather than the road widening it is, MAD and its fellow organizations (CVP and Citizen Marin) can say that Rice has no solutions and is beholden to the madness of us urbanists.
If it’s about politics, then MAD is lying about the project to help their candidate and Haroff is complicit in the deception.
Of course, lots of anti-urbanists see a conspiracy to destroy Marin’s character. They probably actually believe that adding bike lanes is just part of that conspiracy. Al Dugan thinks I work for an anti-Marin lobbying group in DC, for instance.
So some people are playing the political game and don’t care if they’re on the technically correct side of a given issue or not as long as their candidate wins. Others want their candidate to win because they think only by cleaning house can they halt the spread of dangerous ideas in Marin. It’s a potent mix, and it‘s leading Marin down a dark path.
If even a project that will boost traffic capacity by nearly 50 percent is successfully painted as a congestion-causing project, there’s something seriously wrong with our politics. When we can’t even agree on what’s real or not, we cannot have a successful government. Fairfax went through this during Frank Egger’s years on the council. San Anselmo is going through its own turmoil with Ford Greene. Sausalito and Marinwood are going through phases when the whole governing body is dominated by people who take this sort of confrontational and personal approach to governing.
The Drake project has its problems, and I’ve highlighted them before, but as a symbol of our political dysfunction it is extremely worrying.
* For today, I’m not going to get into induced demand and the Fundamental Law of Traffic Congestion, which Connor Jones went over this past winter . That’s a deeper problem, but fixing it is something nobody has the political stomach to take on even in San Francisco, let alone Marin.
** Traffic capacity – how many vehicles a road can carry in an hour – is a function of a road’s design speed and the road’s traffic jam density, or how many vehicles fit per mile when traffic speeds are basically zero:
Maximum Capacity = Number of Lanes × ((Free-flow Speed×Jam Density)/4)
This is called the Greenshields model, which is a reasonable tool for analyzing roads like this. Among other things, it tells us that as speed decreases, the number of cars the road can move per hour actually increases, at least up to a point. This is thanks to the fact that the space between cars decreases as speed decreases, allowing the road to be used more efficiently. Generally speaking, this is around Level of Service (LOS) grade E, though engineers try to keep LOS at around C or D to ensure some slack in the system .
According to this model, reducing the free-flow speed by 3 percent in this model to add a new lane will increase the maximum capacity by roughly 45.5 percent.
 Marin Against Density, MAD Logo, Digital Image, n.d.
 Kentfield Planning Advisory Board, “Sir Francis Drake Boulevard Corridor Rehabilitation” (Kentfield, CA, October 28, 2015).
 Marin Against Density, “Attend June 1 ‘Open House’ -- Forward This to Friends.,” June 1, 2016.
 David Edmondson, “Build Something Better on South Sir Francis Drake,” The Greater Marin, November 18, 2015.
 “Issues,” Kevin Haroff for Supervisor District 2, accessed June 2, 2016.
 Francis Vanek et al., Sustainable Transportation Systems Engineering: Evaluation & Implementation (New York: McGraw-Hill Education, 2014), chap. 7; Wali Memon, “Highway Level of Service,” October 12, 2012.
SMART wants to limit alcohol on its trains to only what patrons buy at the concessionaire. Experience from Caltrain shows that allowing riders to BYOB policy is not just good policy, but helps embed the system further into the culture of riders.
For 18 years I have ridden Caltrain - and in that time I have brought aboard and consumed hundreds of bottles and cans of beers, which I have enjoyed legally on the train. This includes frequent patronage of the semi-official Party Car formed by the cyclists on Caltrain.
Starting in 2000, alcohol consumption on Caltrain increased exponentially with the opening of AT&T park, home of the San Francisco Giants. Giants fans have flocked to the train, riding up the Peninsula with cases of beer and bottles of who knows what, safely being carried to and from the games. At some point, Caltrain decided to ban alcohol on trains running after 9 PM only IF there is an event - primarily Giants games but also Sharks games, concerts at AT&T Park, and now 49ers games and concerts at Levi's Stadium. That late, the consumption before and during the events reach enough of a pitch that it was prudent to put a limit on the policy. Over the years the train has also served hugely alcohol-fueled events like Bay to Breakers, Hardly Strictly Bluegrass, Octoberfests, etc... frequently on the same day.
Generally speaking, this open BYOB policy on Caltrain has been a success. Problems are generally very rare, given the nature of the train as primarily a commuter rail with a higher level fare. It has been an attraction to the train that has a nominal positive influence on overall ridership numbers.
At the end of 2016, I will start riding a new train line - the SMART train in Sonoma County. It will function as primarily commuter rail, running almost exclusively during peak commute hours, with a fare structure prohibitive to general miscreants, making it nominally similar to Caltrain, except that it will serve no special events like the Giants as there are no major sports or entertainment venues on the train line.
SMART has released their draft code of conduct for the train. It includes a policy of NO BYOB. They don't have a no alcohol policy because they have an operating theorem of having a bar car on the train. I am very disappointed by this policy - I find it rider hostile and that it will have a negative impact on the rider experience and overall ridership. While there will be a bar car - there are bar cars on the Amtrak Capitol Corridor too, but in a place like Sonoma County with an excellent selection of beer and wine, to limit riders to the meager selection of a bar car is misguided.
There is, of course, the suspicion that the rationale is not to prevent unruly behavior, but to support whatever vendor they get for their bar car. This is understandable, for the most part because the decision to put a bar car onto the trains is misguided at best. SMART put out a presentation on the bar car where they are toying with giving free rent to the vendor for a return of a percentage of the profits. This is predicated on the presumption that a bar car will turn a profit - my experience from the Amtrak Capitols is that the bar car is at best a loss leader, not a profit center. This is especially true when you consider that SMART has wasted valuable train space to build the bar car.
I personally think that the potential ridership of SMART should make it clear that the no BYOB policy is not in line with the overall goals of the train. They should reconsider this path and allow riders to bring their own beverages onto the train.
If you agree - please email SMART at firstname.lastname@example.org.
Originally Published: Murphy, John. 2016. “SMART Train Alcohol Policy.” Holier than You Blog. May 18.