Tuesday, October 6, 2015

Digging deeper in to Belmont costs

I posted on the very high costs that are driving the poor planning process for the Waverley and Belmont Commuter Rail stations. I've now found the slides from the T's presentation here. Of particular note is page 17, which details costs for the project:

Now let's take a look at some of these numbers. I'm not sure what General Conditions or General Requirements are, although if they're so general, I'm not sure why the are higher for the longer platform.

Structural-01 Platform: $5.5 million for the platforms? That's way too high. Remember, Uphams Corner, Morton Street and South Acton were all rebuilt for between $6.5 and $9 million, with full high levels, and ramps or elevators, landscaping, systems and all else. The cost for the platforms alone—and I'm interpreting this from recent T bid documents—is in the $2 to $3 million range. So there's $2.5 to $3.5 in savings.

Structural-02 Stairs Ramps: Somehow, the ramps and stairs to get to a full-length platform cost $1.5 million more than the access to a mini-high. This makes no sense at all. Since a high platform is higher, and you're accessing it from above, you need fewer stairs and a shorter ramp, so it should cost less.

Structural-03 Elevators: If you have ramps, you don't need elevators. The Fairmount line gets by just fine with ramps and stairs, most of which are just as long as such a facility would require at Waverley or Belmont. There's another $4.3 million saved. (Not to mention lifecycle costs of outdoor elevators.)

Systems: Again, I'm not sure why the systems for a longer platform would cost double the systems for a shorter one. Do you somehow not need lighting and such for the low part of the platform?

Site work, safety controls, track work: These are all relatively small items, but probably all inflated.

Then there are two big ones: Construction Expenses and Professional Services. Each of these, for a full-length, high-level platform, would cost about as much as the entire cost of the full-length, high-level platform stations built in South Acton, Uphams Corner or Morton Street. Each of them—combined, they cost double! It should not cost $7 million to for professional services to plan and engineer a high level platform, nor should it cost that much to construct it.

Good lord, at $100 per hour, that's 70,000 hours—8 person-years, round-the-clock—of work. At 40 hours per week (2000 hours per year), this is enough to pay for a team of 35 people to work on the project for a year to design it. That doesn't make a lick of sense. A full rebuild of Mass Ave in Arlington costs $7 million, but each of these items is $7 million alone, and the set of platforms is nearly as much, with the entire project four times as much? For 1600 feet of platforms, a couple of ramps and some stairs? Come on. These numbers are, as is all too common in Massachusetts, a misuse of taxpayer money, a giveaway to consultants in the place of good planning and cost controls.

How do these costs compare to other cities? Not well. In the Philadelphia area, SEPTA has rebuilt many of their Commuter Rail (well, "Regional Rail") stations with high level platforms. They are somewhat shorter, in the 500 to 600 foot range, but the costs are much lower—in line with what the T spent along the Fairmount Line. For instance, the Fort Washington station was rebuilt for $6 million, including ramps, platforms and a station structure. Not $30 million. $6 million.

Which, of course, is right in the ballpark for what similar station upgrades have cost in Boston. Maybe Waverley will cost more if it is integrated with future development. But $30 million for a single station is so far outside the envelope of reality that it must be questioned.

Here's a quick (and by quick, I mean half an hour quick) sketch of what you could conceivably do with the Waverley Station. This is no more complex than any of the rebuilds on the Fairmount Line, and less complex than South Acton, so it should have costs in line with those projects ($6-$10 million):

This would not interact with any of the retaining walls, which the T claims ratchet up the costs, and would avoid the slight curve and superelevation at the current station site for level boarding. Both ramps would be about 300 feet long—the same length of ramps at similar stations on the Fairmount Line. A small taking would be required from the parking lot of the car wash to the north of the rail right-of-way and east of Trapelo Road. You don't need four elevators for this. You need some concrete. And some actual thinking.

Monday, October 5, 2015

MBTA's Belmont cost estimates are way out of line

Belmont might be getting a new Commuter Rail station. This is not a good thing. Why? Because the T is proposing to create a new parking-oriented station which is within walking distance of far fewer customers than the current stations, and then close those stations.


In 2012, the MBTA performed enough maintenance on the platforms at Waverley Square that it required them to bring the station in to compliance with the Americans with Disabilities act. The problem? The station was built in 1952 when the grade separation there was built (the second-to-last of the B&M's grade separation investments; Winchester was the last) and is in a deep trench, with 20 feet of steps separating the railroad from the roadway.

Now, if a sensible organization were in charge, they'd spec a new station at Waverley accessed by ramps from the street above, full-length, high-level platforms, and build a new station. On the south (inbound) side, there is plenty of room for a ramp from street level to the platform (especially since with a four-foot-high platform, you only need to descend 20 feet). The width of the south side of the station is about 200 feet, so it's a single-zig ramp to get up and down with about three landings on each. The right-of-way, which held three tracks until the grade separation, is 100 feet wide, so the north side (outbound) platform could be accessed by a ramp to the east of the station to a rebuilt platform, with the current stairwell providing access from the narrow part of the triangle. Alternatively, platforms could be moved to the east of the station where the line is straighter, and the current station area used for vertical circulation.

In Belmont Center, the solution is easier. It would be hard to build new platforms at the current station location, which is located on a curve and would necessitate a Yawkey-style double-platform solution: a costly endeavor. However, just east of station the tracks straighten out and the right-of-way is 84 feet wide, plenty for a pair of platforms and two tracks. The north side of the station would be accessible with a ramp in place of a short staircase and a short ramp to a high level platform, which could be build to the east of the current platform. The south side would require a longer series of ramps, probably just east of the station. But again, this is not insurmountable.

But what does the T want to do? The T wants to close both stops and build a new station in between. This new station would be within walking distance of fewer residents, and much further from the downtown employment areas at Belmont and Waverley (as well as McLean Hospital, the largest employer in the town). It would have a parking lot, so a lot of people who currently walk to the train station would have to drive. It would sever connections between the bus lines which serve Belmont and Waverley and the train station. (Do you live in Waltham and work at Mount Auburn Hospital? Say goodbye to your rail-bus commute, you'll have to take the train in to Porter, the Red Line to Harvard, and the bus out.)

Does this make any sense? If you're a Belmont resident, it does not. Apparently, if you're the MBTA, it does.

Here is a quick table of the current stations and their catchment areas (defined as the total population of all census blocks all of part of which are within 0.5 miles of the station location):
Belmont (current): 4311
Waverley (current): 7452
Proposed location: 7019
So not only does the proposed location serve far fewer residents than the current two stations, it actually serves fewer residents than the currently Waverley Square station!

How does the T justify this farce of a plan? The same way the justified killing the Red-Blue connector: by wildly inflating the cost of the plan they don't want (bringing the current stations in to compliance) and giving few if any details about the costs of what they want to build. Their estimated cost for rebuilding each of the Belmont stations? $35 million (well, they've now come down to $16 to $30 million). Their number for a new station? They can't say, but think it will be less. This is so entirely disingenuous it knocks my socks off.

Let's look at some data to see how far off these numbers are, looking at some recent station construction for MBTA stations with full, high-level platforms and note whether it was a rebuild or a new station and the type of vertical circulation:

Yawkey Station (Worcester Line), 2014: $13.5 million
Existing station. built on a curve, in a narrow corridor, required track realignment, and elevators.

Boston Landing (Worcester Line), 2016, $20 million
New Station, ramps

Newmarket Station (Fairmount Line), 2013, $12.3 million
New Station, ramps.

Uphams Corner (Fairmount Line), 2007, $7 million
Existing station, ramps.

Four Corners/Geneva (Fairmount Line), 2013, $17.7 million
New station, ramps.

Talbot Ave (Fairmount Line), 2013, $15.9 million
New station, project cost included a new bridge, ramps.

Morton Street (Fairmount Line), 2006, $6.5 million
Existing station, ramps.

Blue Hill Ave (Fairmount Line), 2017, $10 million
New station, center platform, ramps.

South Acton (Fitchburg Line), 2015, $9.6 million
Existing station, elevators. The T is willing to flat-out lie about the construction costs by including land acquisition costs, which are not an issue for either station here, considering the 100-foot right-of-way.

Littleton (Fitchburg Line), 2014, $8 million
Existing station, center platform ramps.

So let's review. Existing stations upgraded to full high level platforms built in the last 10 years have ranged in cost from $6.5 million to $13.5 million, with an average cost of $8.9 million. New stations built with full high levels have ranged from $10 million to $20 million, with an average cost of $15.1 million (the ones with elevators cost more, and elevators have higher maintenance costs). So the T is being disingenuous in two dimensions. First, upgrading the two existing stations, based on these averages, would cost only slightly more than building a new station. And secondly, recent experience shows the price would be nowhere near $35 million dollars for each station. Not even in the ballpark!

In fact, the T's own planning documents admit this and quotes a cost of $15–$18 million for stations with side platforms (double the cost of recent station rebuilds, but still nowhere near the $35 million figure) and $22–$30 for (new) center platform stations (based on the cost of the Boston Landing station, although Blue Hill Ave is center platform and half the cost of Boston Landing). So while this planning document inflates the cost of upgrading stations, it still doesn't get it to the $35 million level. And the claims that new stations cost more than rebuilding existing stations are just wrong.

Now, the T will claim that the new station is needed because the current stations are underutilized. This is hokum. The current stations are underutilized for two very good reasons: frequency and fare policy. First, frequency:
Waverley: 9 departures, 3 peak, 2 reverse commute, 4 off-peak
Belmont: 14 departures, 4 peak, 2 reverse commute, 8 off-peak
Waverley: 12 departures, 4 peak,  2 reverse commute, 6 off-peak
Belmont: 13 departures, 4 peak, 2 reverse commute, 7 off-peak
If you're in Waverley and you don't need to get to North Station at a very specific time, it's better to take the 73 bus to Harvard and the Red Line. The 19 minute travel time is half of the bus-subway route, but it only runs once an hour. In Belmont, you have a few more choices and fewer buses, but there's still no frequency. Schedules are uneven, and some midday trains skip the stations. Departures should be rationalized once the line has full double-tracking, but the stations won't attract more passenger without more service. Building a different station won't help.

Then there's the fares. A trip on the bus and subway from Belmont to downtown costs you $2.10, or $75 for a monthly pass. A train ticket? That comes at $5.75, or $182 for the month, nearly triple the cost. This is a longer discussion about T fare policy, but it doesn't help drive ridership in Belmont. Still, neither is an argument to build a station which will serve fewer people. (And, no, a new station won't be an express stop: the "super express" from South Acton doesn't even stop at Waltham.)

So what's going on here? I'm not sure if it's a question of incompetence or something more nefarious. But it seems to be in line with a lot of what goes on with a lot of Massachusetts infrastructure projects: create an agenda (single station in Belmont), create a plan that is unpalatable to the community (less access, more traffic) and make up costs so that your agenda compares so favorably that it is the only logical option. Then advance it far enough that it's illogical to turn back. We saw this with Red-Blue connector, which has cost estimates the Globe has called "deliberately high." We've seen it in Allston, where this page has argued that the state is pursuing a far more expensive plan by being immovable on project guidelines and refusing to fully vet better alternatives (although there is some progress being made). Unfortunately, this Belmont "plan" fits the pattern all too well.

What should we do? Contact your legislator, especially if you live in or near Belmont. Go to meetings. Raise your voice. And let the T know that they can't get away with further degrading Commuter Rail service by making up numbers.

Part 2 takes a deeper look at these numbers, and finds them to be no less made up.

Thursday, September 17, 2015

Allston presentation link

A lot of people have asked for the Allston presentation that I gave to the Task Force. It's here. Enjoy (?).

Update 9/28: Here is the "dueling" presentation from ABC. Dueling very much in quotes, because I think they are 90%+ the same. I've worked a lot with the authors, and think both have merits. (You may even see a slide in there from me :)

Wednesday, September 2, 2015

Something is very wrong with MBTA project procurement

When the news broke about the escalating costs of the MBTA's Green Line extension to Medford and Somerville, I can't say I was surprised, but the cost numbers have now escalated in to absurdity. I'm not an expert in the bizarre project procurement, but the costs are now to the point where the project really should be reviewed and rebid, even if there is a delay. It is far beyond what similar projects cost in other cities, and the project procurement team at the T should be removed to allow someone from the outside to bid the project.

How ridiculous are the T's numbers? Let's take a look at them, and compare them with some other projects:

Minneapolis-Saint Paul Green Line
Distance: 11 miles
Cost: $957 million
Cost per mile: $87 million
Completion date: 2014
Engineering difficulties: rebuilding bridge over Mississippi River for light rail, rebuilding the entirety of University Ave lot line-to-lot line, junction and flyover with existing Blue Line in Minneapolis. Likely cheaper than the Green Line extension. But not 9 times cheaper.

Los Angeles Expo Line Phase II
Distance: 6.6 miles
Cost: $1.5 billion
Cost per mile: $227 million
Completion date: 2016
Engineering difficulties: Several grade separations, parallel bike/walk facility. Still four times cheaper than the GLX.

San Francisco Central Subway
Distance: 1.7 miles
Cost: $1.6 billion
Cost per mile: $941 million
Completion date: 2019
Engineering difficulties: full deep bore tunnel in a seismically active area with three underground stations. In a rather expensive city to work in. And barely more than the GLX, which is being constructed in a grade-separated right of way!

Seattle University Link
Distance: 3.1 miles
Cost: $1.9 billion
Cost per mile: $613 million
Completion date: 2016
Engineering difficulties: full deep bore tunnel below the water table in a seismically active area with two underground stations. And quite a bit cheaper per mile than GLX.

Now, here are two MBTA projects:

Red-Blue Connector
Distance: 1500 feet
Cost: $750 million
Cost per mile: $2.6 billion
Completion date: ???
Engineering difficulties: Cut and cover tunneling below the water table in a constrained corridor. Certainly no greater than building the Central Subway in San Francisco, yet somehow three times more expensive. This should probably be in the lower end of the $100 to $200 million range, not three quarters of a billion.

Green Line Extension
Distance: 4.3 miles
Cost: $3 billion
Cost per mile $700 million
Completion date: 2019?
Engineering difficulties: Relocating existing parallel commuter rail line, building a flying junction, parallel bike/walk facility.

Here's the thing: none of the engineering challenges faced by the GLX and RBC are unique (flying junction in Minneapolis, parallel path in LA) or insurmountable, yet the costs are an order of magnitude greater than in other cities. The Green Line Extension is between three and nine (!) times more expensive than similar light rail lines, and more expensive than new light rail lines which are being built using deep bore tunneling techniques, which are not cheap. High construction costs? Seattle and San Francisco have pretty high construction costs and labor wages too. The remaining GLX construction should be rebid mimicking the processes used in these cities with a new team at MassDOT, and if costs aren't cleaved significantly, there should be a full investigation as to why.

The for the Red-Blue connector, which everyone agrees is a very important link, somehow costs three to four times what much more complex projects cost in Seattle and Los Angeles (while the Blue Line uses heavy rail equipment, it is the same diameter as light rail trains). The project is only 1500 feet long, doesn't require a deep bored tunnel, and has only one station at Charles, and the headhouse there already exists with provisions to connect it to the Blue Line. The fact that it costs $2.6 billion dollars per mile is laughable. For the cost of one mile of construction in Boston, Minneapolis could build 30 (!) miles of light rail, and Seattle four miles of deep bore tunnel (about what you'd need for the North South Rail Link) and at a cheaper rate per mile than the Green Line Extension. It's not even in the ballpark of reality, and whoever at MassDOT comes up with these numbers needs to be sent out to pasture.

There is no logical reason why a project in Boston should cost triple—or more—what a similar project costs in another city. $2 billion was suspiciously high. $3 billion means that a lot of people are on the take, or that money is being pissed away. I'm all for transit expansion, but not at these prices. An outside manager is a good start. But this has been a problem for a long time, and the T's project procurement staff has shown no ability to do their jobs. Get rid of them.

This is correctable. It needs to be corrected.

Tuesday, September 1, 2015

Quick thoughts on the ½ block Mass Ave bike lane

I took the liberty to bike home this morning from running across the Harvard Bridge to check out the new bike lane there (sans phone, however, so no sexy pictures). Here are some quick observations:

  • It needs flexi-posts. Too many drivers are used to using it as a right turn lane. Once installed, they should preclude that, which is where the real safety improvements come (that and No Turn on Red markings, which are long overdue; perhaps DCR and MassDOT will match with NTOR on the Cambridge side of the bridge). Whatever system is put in place needs to be kept year-round, with a snow-clearance plan (which includes the bridge, which MassDOT did not clear this past winter).
  • Flexi-posts would be nice on to the bridge, which is always a bit harrowing with catch basins on the right and traffic on the left, but that would mean it would be hard to pass slower cyclists, a relatively frequent occurrence. Perhaps intermittent (every 30 meters/Smoots) posts that would still allow cyclists to change lanes would work. It's also MassDOT territory and interagency cooperation is not a thing in Massachusetts.
  • Once in place, there will no longer be the merge/bus stop/right turn hell that has been the situation there for years. (Or as vehicular cyclists would call it: paradise. To be dead serious for a moment: the previous situation at Mass and Beacon was exactly what vehicular cyclists advocate for: a free-for-all with bikes acting as cars. People died because of it. It is high time for the vehicular cyclists to be banished to the dustbin of history.)
  • The bike lane westbound/northbound on Mass Ave (on the other side) has been striped quite wide—at least 8 feet—coming in to Beacon, where it shares a bus stop. This means that both travel lanes are just nine feet wide. If nine foot lanes are okay (and they should be) you could dramatically improve Mass Ave.
  • The current lane off the bridge is probably the most important single block for a bicycle facility in the City of Boston, and perhaps the Commonwealth. It is heavily used, with high conflict rates and a steep downhill off the bridge which sent cyclists flying in to a sea of cars and bad pavement. So it's the right place to start. But it's just a start. The lane really needs to be extended all the way down Mass Ave (as has been proposed here and elsewhere). 
  • By moving the bus stop to Marlborough, a few parking spaces will be lost. And the M2 will have to find a new stop (I'm not sure where). In fact, it might make sense for both bus stops to be moved to either side of the Marlborough intersection, which would make my previous plan even more feasible (expect a redraft of that in the next couple of days).
  • Still, I think that this is a huge change, because the city eliminated a lane of traffic without a months-long, drawn-out "traffic study" which would claim a reduction in LOS and that therefore a bicycle facility couldn't be accommodated. Is it reactive instead of proactive? Yes. And that needs to change. But it sets a precedent: we can remove traffic lanes in the name of safety. And it's now time to act for a safer Mass Ave, a safer Beacon Street, and safer streets all around.

Friday, August 28, 2015

Poor transit planning: an example that proves the rule

One of the most infuriating things about transit planning is when an agency does something which is designed to cause delays and for which there is an obvious solution. Customers have some understanding about things that are beyond the railroad's direct control: trespassing, downed trees, even (to an extent) equipment, track or signal failures (although these are more preventable). But when your bus or train takes longer than necessary because of a stupid decision by the transit operator, it is far more maddening. These would include things like fare policy (the front-door boarding on the Green Line, which causes completely unnecessary delays for passengers). But that comes in relatively small doses, and it's hard to compute.

A far more concrete example is what happens late every morning on the Haverhill Line. Starting this week, all midday service between Haverhill and Boston has been sent via the Lowell Line to allow for track work on the southern section of the line. What the T did is to take all of the scheduled Haverhill trains and run them express via the Wildcat between Ballardvale and North Station. They maintain their original arrival times, but by running express via faster trackage should arrive at North Station 15 minutes early, so the arrival time is schedule padding. Except in the case of one train, 210, which will arrive late every day, because of a scheduled conflicting move.

210 leaves Haverhill at 9:05 and is scheduled in to Ballardvale at 9:29 before its express trip to Boston. Amtrak's first northbound Downeaster leaves Boston at 9:05, Woburn at 9:23 and Haverhill at 9:53. During normal operations, this works fine: 210 clears Wilmington Junction staying on the Western Route (the track towards North Wilmington) around 9:30, and Amtrak approaches northbound on the Wildcat a few minutes later. With the schedule change, however, Amtrak takes precedence on the extended single track, from Wilmington all the way past Andover, a distance of 8 miles (there is a double-track project in the area which will mitigate this issue; it is proceeding at a decidedly glacial pace). So the commuter train has to sit there for 15 to 20 minutes waiting for Amtrak to pass. Every single day.

But there's an easy fix to this: the commuter train could leave Haverhill 15 minutes later. There are no schedule conflicts, no track conflicts: the train runs 15 minutes later, approaches the single track at Andover just as the northbound Downeaster gets on to the double track, and proceeds in to Boston. Everyone on the train has a ride that's 15 minutes shorter. Everyone is happy.

I'm sure the argument against this would be something like "people are used to the train showing up at a certain time and we don't want to change the schedule on them." This is why you have a schedule change. The train would be moved later. A few people who don't pay any attention will wind up waiting 15 minutes at their station the first day. Some of them might consult a schedule posted there and realize that the schedule was changed, and not come 15 minutes later the next day. No one will miss the train, only a few people would have a longer wait. Which they are going to have anyway, because nine days out of ten the train is going to be held for conflicting traffic (unless the Downeaster is delayed significantly, and it's been running on schedule more frequently recently).

Plus, the current delays will lull passengers at Andover and Ballardvale in to thinking the train will always be 15 to 20 minutes late, so they'll start showing up then anyway (Pavlov's dog, etc). Then, on the one day that the track is open and the train operates on schedule, people will be left behind (at least 212 operates rather shortly thereafter). So by not changing the schedule to reflect reality, the following occurs:
  • All passengers from Lawrence north have a 15 to 20 minute on-train delay every day north of Andover.
  • All passengers at Andover and Ballardvale have a 15 to 20 minute on-platform delay.
  • On the occasion that the train operates on schedule, any passengers who decide to come at the actual usual arrival time at Andover and Ballardvale run the risk of missing the train, although they may decide that the 15 minutes saved most days make up for the 40 minute delay once a week.
Pushing the departures back 15 minutes would solve all of these problems, and give passengers a very fast trip in to Boston express from Ballardvale. So why doesn't this happen? Perhaps institutional inertia. Perhaps incompetence. Maybe the T doesn't even want people to get used to a faster trip when the trains are shifted back to the local schedule in a few months. But it's a sad state of affairs when the T and Keolis decide that a 20 minute delay is acceptable—every day—and the passengers are the ones who pay. 

And then they wonder why the passengers lose all faith in the transit providers. This is why.

Wednesday, August 26, 2015

Burying the lede: Caltrain saves money vs driving, not Uber

There's an article about a woman who replaced her car (well, SUV) with Uber and how much money she saved. A lot of the variable costs were from her commute "from home to KP in Menlo Park was 70 miles round trip and I’d go down to the office 2–3 times a week." That's quite the commute to do by Uber.

She spent $4700 on Uber last year.

The cost of an Uber from Sand Hill to San Francisco is $51 to $68, without any surge. An average roundtrip would cost $120. So if you did this two to three times per week for a year it would cost $15,000. It's pretty clear that she's not replacing Uber trips with car trips, or she wouldn't have made it to May. So without a car, how does she get to Menlo Park?

By train, of course. Caltrain runs a nice service south from the City, with frequent trains, many of them express. She uses Uber on either end (you know, there are these bike things, and buses, but those are hard). So there's probably an $8 Uber ride on either end, and the other $104 are on Caltrain (40 minutes, generally faster than traffic). Except it costs $13.50. So the whole commute costs $30, or ¼ of the cost of riding Uber both ways, all of the savings from transit.

Taking Uber both ways would cost 50% more than driving. So that doesn't work. What she really means to be saying is "I started taking transit for the majority of my commuting miles and saved a ton of money." The Uber use helps the transit work in this case, but it's Caltrain that is the main reason this works, not just Uber.