Wednesday, June 25, 2014

Common sense MBTA Commuter Rail improvements

With the MBTA dead set on spending 2.4 billion dollars to bring choo-choos to Fall River and New Bedford, I thought I'd give the a few suggestions for what to blow their next pile of money on. The difference is, my improvements bring riders, not suburban votes.
Electrification: Worcester, Fairmount and Providence
This is key for both its relatively good effects and its extreme low cost.  The primary cost is strining up the wires on the Worcester and Fairmount lines, done on the NEC recently for 5 million per mile, plus the cost of ordering electric locomotives to pull the trains, recently ordered by Amtrak for roughly 5 million per unit. In exchange, we get much better trip times for both Worcester and Providence, and due to the frequency expansion possible on the Fairmount (see previous post), amazing ridership increases. On the Providence line, it also allows the T's trains to "get out of the way" of Amtrak's services, possibly increasing capacity.
DMUs and frequency: Everywhere
Frequency is key, and I cannot stress enough how much of a game changer higher frequency can be. With European style Diesel Multiple Units, like the ones currently being considered for the Fairmount line (EMUs would be better, but more expensive), higher frequencies are achievable with fewer cars and, more importantly, fewer operators. Therefore, the costs of this improvement are, if handled correctly, offset by its savings. The sole cost will be the few tens of million to order the DMUs, which is a bargain compared to any sort of rapid transit expansion.
Take the Worcester line into the 20th century: Worcester
The Worcester line currently takes 90 minutes go to 45 miles, on an express train. The local takes a painful two hours, and runs with a frequency that makes your average commuter rail line look like rapid transit. For a few hundred million dollars in additional track, plus several million for a new signal system, the lines performance could significantly improve, probably down to a 70 minute express. When combines with electric service from suggestion one, this would result in drastically improved ridership. Thanks to the almost-intercity characteristics of the line (large cities at both ends), frequency is less important than speed, so hour or half-hour frequencies all day should be acceptable.
Any suggestions? Post them below!

Sunday, June 22, 2014

Amtrak's new Acelas, and high speed rail in the northeast

As some of you may have heard, several months ago Amtrak and CHSRA, the organisation responsibly for the development of High Speed Rail in California, decided to put out an RFP, a request fr proposals, on a  new high speed train to replace the existing Acelas and start high speed rail service in CA. Now, according to the Fresno Bee, they're cancelling the RFP and starting over, with the two organizations putting out two different sets of specs.
Why this happened:
I'm actually surprised this didn't happen sooner, for several reasons. The first involves the required performance characteristics of both routes. Acela IIs must travel at 160 MPH over twist track, and they do this via a mechanism called "tilting", where the train rotates the passenger cars to lessen lateral forces around a curve. Because of the extreme curvature of the Northeast Corridor, this is a must to achieve the improved trip times over ordinary Northeast Regionals. In California, however, these trains will run over a completely new, world class right of way at 220 MPH, and will have no need for tilt due to the straightness of the alignment chosen. Choosing to issue a joint RFP was contingent on manufacturers being able to deliver a trainset that met both specifications. As we learned recently, none can. The next step will be to release new RFPs, one for both agencies, detailing only what the agency wants, and wait for responses. This should push back delivery several months, and service is now expected to begin late 2018 for Amtrak. Even that date is seriously optimistic, considering it took ten years from RFP to revenue service the first time.
Why these trainsets are being ordered:
I've heard plenty of people asking a perfectly reasonable question: why are these trainsets being ordered? The existing trains are only 15 years old, and have reliability and OTP numbers that exceed every other Amtrak service. The short answer is there isn't enough Acela to go around. With only 300 seats per trainset (the cars are semi-permanently coupled and and cannot be lengthened), and only 20 trainsets, the booming demand for rail travel in the Northeast has caused heavy crowding, and especially south of New York, frequent sellouts. This has, as anyone who's priced out a trip knows, resulted in ridiculously high fares.
How they'll improve service:
Thanks to the failed joint RFP, we have a pretty good idea of what the trainsets will look like. They'll do something unusual in the US; they'll be EMUs. An EMU, or Electric Multiple Unit, is a train with motors and powered axles under every car, with no locomotive. In addition, passengers sit in the front and back car, currently taken by the locomotives, behind a cockpit style engineer control panel. By putting all the traction guts under the floor of the car, the entire train can be turned into seating, increasing capacity. Pretty much every other high speed rail system in the world, including CAs, will use EMUs for exactly that reason. EMUs also offer advantages in acceleration, especially if Amtrak can get out of the FRA's inane requirements involving crash strength (this will be the subject of another post). In total, we'll probably get five or ten-minute trip time improvements Boston-New York, and lower prices due to capacity expansion.

Saturday, June 21, 2014

BRT versus parking, and making hard choices about street space

What's the most valuable commodity in Cambridge? It isn't water, open space or even housing. In fact, it's street space. Whether given to parked cars, bikes or  pedestrians, it's worth its weight in gold. Therefore, in urban areas like Cambridge, we need to be as efficient as possible about how we use that space. This post is dedicated to talking about the hard choices involved in allocating a given road's surface. For my main example, I'll use Mass Ave south of Porter Square as an example.
Current built environment:
Mass Ave south of Porter currently consists of a separated boulevard with an incredibly thin concrete median. On either side there exists two general use lanes, each roughly twelve feet wide, and one continuous parking lane (no curb bump outs), taking about 8 feet of space. It hosts several bus routes, one, the 77, being a Key Bus Route. There is an incredibly temporal bike lane on the northbound side, and there are 8 foot sidewalks, with frequent obstructions, at the edges.
First, lets see how much space is given to any given mode. I'll use a winner take all system: if a lane is dominated by one type of traffic, I'll give it all to that type. Therefore, general purpose lanes, which are over 90% car, will be counted as car lanes.
Doing the math on street space
8(Sidewalk)+8(Parking)+12(Car)+12(Car) :Median: 12(Car)+12(Car)+4(Bike)+8(Parking)+8(Sidewalk)
So, out of an 84 foot road surface:
64 feet are given to cars,
16 feet are given to pedestrians, with frequent interruptions,
an insignificant 4 feet are grudgingly given to bikers,
and a grand total of 0 feet is given to transit!
To do the percentages, over 75% of the road is given to cars, in the form of parking and travel lanes.
So much for complete streets!
That's a pretty grim picture. Even after being given 64 feet of space on the most valuable road in Cambridge, cars are constantly congested. They move slower than bikes, despite being given 16 times more space! In short, cars are a miserable failure at actually getting people where they need to go in an urban setting.
What we can do:
Therefore, it's time to rethink how we design roads. Lets start by taking two of the car's six lanes on Mass Ave (remember who parking is for), and dedicating it to buses, which have proven themselves much more efficient per square foot of urban space. Then, lets take the parking out, which has a capacity of exactly zero bodies per hour (I will do many additional posts on the folly of on street parking), and replace it with well designed cycletracks. Finally, to decrease speeds and increase safety, lets narrow the two remaining car lanes to 10'6", the minimum allowed for emergency vehicle access, and give the reclaimed three feet to pedestrians in the form of wider sidewalks. Now, we have a real complete street, that gives equal space to every mode.
What it looks like now:
Out of an 84 foot road surface:
24 feet are given to transit,
21 feet are given to cars,
19 feet is granted to pedestrians,
and 12 feet is given to cyclists.
That's a Mass Ave everyone can enjoy.

Friday, June 20, 2014

Bus Rapid Transit, and the most cost effective way to stimulate ridership

We will never build another mile of heavy rail in Cambridge. It's over. Besides the quarter mile of new tunnel required to bring the Red Line to Arlington, which is possible on a multi-decade timescale, there is no reasonable place where heavy rail could even be built, let alone be done in a manner that's fiscally prudent. Does that mean new, high speed transit in Cambridge cannot be built? Hell no; it means we need methods with less surface disruption, lower costs and smaller minimum ridership numbers. There are two applicable answers, in the form of Light Rail and Bus Rapid Transit. I happen to be a huge supporter of both, but Cambridge is nearly a textbook example of where buses, previously the lowest method of transportation, can thrive. All you need to do is give them the space to fly (or drive, as the case may be).
What it is: Bus Rapid Transit
Bus Rapid Transit, henceforth referred to as BRT, is the concept of turning street running, mixed traffic buses into something closely resembling Boston's Green line. The central concept in BRT, which is NOT obeyed in Boston's Silver Line, is the concept of dedicated lanes. Why should a bus with 80 riders be forced to stay behind a car with one? By taking the "high speed" lanes in the middle of roads like Atlantic Avenue or Mass Ave, you can *add capacity*, by replacing space inefficient cars with massively space efficient buses. In addition, you create a transit line for the cost of a few buckets of paint. It won't be as fast as the Red Line, but with priority at traffic signals (something the Green Line lacks, for reasons that baffle me), it can get pretty close.
The biggest benefit: cost
For various reasons, all transit in America is expensive. Instead of wishing for a land of $50m/mile subways (It's current an order of magnitude above that), lets work with what we have now and create 5m/mile busways. 
5 million per mile?
Yes, you read that correctly. For 5 million dollars a mile, we can create rapid transit with similar speeds and frequencies to Boston's Green Line. With costs like that, the equation changes from "Why?" to "Why not?". Why shouldn't we turn all of the 66, 1 and 77 to BRT? Why shouldn't we have dedicated bus lanes all over Harvard and Kendall? Why do we dedicate 80sf of street space to one car when we could give it to 20, 30, 40 people on a bus? And who cares about Red Line reliability when you have five transit routes running Boston -> Cambridge, all with 5 minute headways?
The catch:
The answer is, unfortunately, a political one. It order to get lanes for thousands of people on buses, you need to take it away from  hundreds of people in cars, and those users are highly resistant to giving up lane space. It's worth noting that there is a common fallacy involved here: I won't stop driving my car, so traffic will just get worse with fewer lanes. A correct response, though not exactly an obvious one, is that even if you will never switch away from your car to the new, faster bus, someone else will. This solves the problem of congestion on roadways, even with fewer car lanes. In Boston, these challenges have defeated all useful BRT proposals, but in Cambridge, with the master plan on a large number of brains, I'm hopeful we can see the math behind BRT and go on an all-out war for better transit, cheaply.

Thursday, June 19, 2014

Why an Alewife commuter rail stop is a mediocre idea, and a plea for frequency

       In several recent conversations involving Cambridge's developing master plan, I've heard references to an Alewife commuter rail stop as a possible solution for the highly car dependent Alewife area. I think it's a decent idea, especially when combined with a pedestrian overpass, but the current MBTA commuter rail operating practices make it unlikely to generate significant ridership. Fundamentally, it comes down the the concept of frequency, and what makes transit useful.

Frequency: The great decider
When it comes to what makes people ride transit, one of the most common perceptions is that the service provided, in terms of physical infrastructure, determines ridership. For example, in a situation where a light rail line replaced a bus, ridership can be assumed to rise. This perception is partially correct, in that a totally equal service provided with Bus Rapid Transit and full Rapid Transit will experience higher ridership with the steel wheel version. However, there is a far more important characteristic determining ridership, which is essential to understand the poor performance of commuter rail in the United States.
First: Commuter rail ridership
In order to make my argument, it's useful to look at ridership figures for the commuter rail, using the MBTA's invaluable Blue Book. In particular, lets look at Porter Square, which is an "Outside the core" station with both commuter rail and subway service, in addition to several bus routes. If the hardware behind the service determines ridership, we should reasonably expect to see the comfortable, speedy commuter rail in first, more than one thousand riders spread over multiple routes. However, in reality the opposite occurs. By far the most popular mode is the subway, followed in a distant second by buses. In an abysmal third comes the commuter rail, with an absurdly small 250 inbound boardings.
What's happening: Frequency
With these numbers in hand, it would be useful to find a new new deciding factor, one that would predict Subway->Bus->Commuter Rail at Porter, rather than Commuter Rail->Subway->Bus. It comes in the form of frequency. Instead of looking at speed, we need to look at time to access: how long one person needs to wait before transportation arrives, whether it's in the form of a train, bus or rickshaw. In addition, we should look at service consistency: the knowledge that at any time of day, that person could show up and expect transportation quickly. A subway is the ideal model of this, and commuter rail does heinously. When you show up at a Red Line station, you know a train is coming in a maximum of ten minutes (unless the MBTA has discovered a new way to break down, but I digress), and importantly, you know that will be true all day. In stark contrast, the commuter rail has no such guarantee. If you arrive in rush hour you may have to wait "only" twenty minutes, but if you arrive in the middle of the day, you may have to wait for up to two hours. If you need to get somewhere in a hurry, the commuter rail is exactly what you don't want: long waits that you cant predict, short of memorizing an entire timetable. This creates a situation where only 250 people decide to ride the commuter rail, where at least 12,000 choose to ride more frequent, yet less comfortable, modes.
What this means for Alewife:
So what does this mean for Alewife? The short answer is, don't expect great, or even good, ridership for any new commuter rail stops. The long answer also includes a hint for what we should do: increasing frequency, even without superior modes, produces disproportionate increases in ridership. Therefore, a better idea to increase transit ridership at Alewife is perhaps the most basic: better frequency, measured by the mean time between a bus departing a stop and the next one pulling in. For frequency, a great general rule can be described by who a given headway is useful for.
1 hour: Useful for almost no one
30 minutes: Useful for students and the poor
15 minutes: Useful for those without a car
10 minutes: Causes people to leave the car at home
5 minutes: Causes people to leave the car at the dealership, unsold
You'll notice the most dramatic effect occurs at five minutes, exactly the Red Line's average frequency.