Finance Light Rail Norfolk

How significant an opportunity for reducing U.S. construction costs?

» Norfolk, Virginia celebrates the opening of a relatively cheap new rail corridor. It’s not as out-of-the-ordinary as we might hope, though.

Last weekend, Norfolk’s Tide light rail line opened to big crowds and lots of excitement in a state that has never before seen modern light rail technology in action. But the project was overbudget and the subject of years of controversy. What was once supposed to be a $232 million line had ballooned in cost to $318.5 million and in the process taken down political leaders who had supported it. Perceived mismanagement delayed consideration of extensions into nearby Virginia Beach. And the scheme’s implementation flaws emboldened conservative activists insistant on playing up the poor performance of government.

The irony of the story, it turns out, is that even at its higher-than-expected cost, the Tide’s construction came in at just $43 million a mile, less than any recently completed or under construction light rail system in the United States — even better than Salt Lake City’s just-finished Mid-Jordan and West Valley light rail lines, which cost $50 and $73 million per mile, respectively. It came close in price to the cheapest such project in generally less expensive France, Besançon’s $35 million a mile tramway.

What does this mean for other U.S. cities hoping to keep costs down and get the biggest bang for their buck? Has Norfolk done something tremendously different than other places?

One explanation is the corridor chosen for the rail line: The 7.4-mile right-of-way was almost entirely located within either an existing freight railroad corridor or along public streets. This reduced the need for land acquisition and land grading. It also made it simple to serve some of the region’s most important destinations pretty directly, including the Medical Center west of downtown, Norfolk’s central business district, a baseball stadium, and Norfolk State University. The line’s position along the edge of I-264, a major highway, and the quite limited residential and commercial populations of downtown Norfolk won’t do the project any favors; its eleven stations are only expected to attract 2,900 daily riders this year and 7,200 by 2030. But its alignment parallel to Virginia Beach Boulevard — the route of the most popular bus line offered by Hampton Roads Transit — means it would likely do better than any other rail line in the city.

Let’s return to the matter at hand, though: Just how different is Norfolk? Has it been able to apply some magic elixir to reduce its costs of construction?

Perhaps the most accurate answer is that it’s not clear. A review of 32 urban rail transit projects across the nation that have either been recently completed, are under construction, or are soon to enter construction suggests that there are limited margins for cost differentiation among similar projects.

In the following chart and the table at the conclusion of the article, I have compared service miles with construction costs among five types of projects — some that are fully underground; some partially underground, partially above ground; one that is fully elevated; some fully on the surface (or partially elevated); and two fully on the surface with only one track in service for parts of their routes.

As the chart demonstrates, there is clear evidence that the type of service provided — surface, elevated, or subway — is the primary determinant of construction cost differences. Unsurprisingly, of course, rail projects that operate within independent rights-of-way such as along elevated viaducts or underground are likely to cost more than similar-length projects running at ground level. The easiest way to lower construction costs would be to convert every subway project to an elevated and every elevated line to surface running and every surface-running corridor to one with just one track.* Yet this is no answer at all: This would be an ineffective solution, since it would reduce capacity in corridors where it is necessary.

But within each of these groups, evaluating like project to like project, we are given the opportunity to compare similar schemes addressing similar ridership demands. If Norfolk is cheaper, why? How do light rail programs serving equivalent populations fare?

Unfortunately, this comparison indicates that there isn’t much we can do to differentiate between projects. For surface-running lines and those that are planned to run on the surface and in subways, for instance there is relatively strong evidence (note those trendlines) that construction costs will be close to $73 or $239 million per mile, respectively.

The most expensive surface-running rail line now under construction is Portland’s Portland-Milwaukie light rail, which will cost $204 million per mile to construct as of the most recent estimates. That’s expensive, but it includes a significant bridge over the Willamette River and a series of elevated sections. The large majority of light rail lines like Norfolk’s will come in at less than $100 million per mile, most between $50 and $70 million per mile. Streetcar lines, running in shared automobile lanes, are a bit cheaper. Overall, this suggests that regional differences do not seem to matter much (see Sacramento’s South Corridor, in a union-friendly state, versus Phoenix’s Central Mesa Extension, in a right-to-work state), nor a reliance on federal funding (see Salt Lake’s West Valley line, which was funded without Washington’s support), nor indeed the existence of private investment (see Denver’s Eagle project, funding that city’s East and Gold Lines).

What conclusions can we take from these data? One must be that construction costs in the U.S. are relatively steady across the country, at least when taking into account differences in grade separation. The other is that if we consider it in the public interest to reduce construction costs because of a declining ability to afford infrastructure, a national solution, rather than a local one, may be necessary.

Table of recent American rail transit projects, either under construction or soon to enter construction
CityProjectProject TypeCost $mMilesCost $m/MileRiders 2030Cost $/Rider-Mile
St. LouisDelmar Loop TrolleySurface, 1 Track442.22026007692
New OrleansUPT StreetcarSurface451.530
NorfolkTide LRTSurface3197.44371006072
SeattleFirst Hill StreetcarSurface1252.55090005556
Salt LakeMid-Jordan LRTSurface53510.650190002656
Salt LakeDraper LRTSurface2063.85468007972
DenverWest LRTSurface71012.159
SacramentoSouth II LRTSurface2704.363100006279
PhoenixCentral Mesa LRTSurface1983.164
DallasGreen/Orange LRTSurface14062167459001459
DenverEast/Gold LRTSurface204330.268575001177
Salt LakeWest Valley LRTSurface3705.173105006909
Twin CitiesCentral LRTSurface9571187409002127
CharlotteNortheast LRTSurface118011107175006130
HoustonSoutheast LRTSurface8236.5127283004474
HoustonNorth LRTSurface7565.2145282005155
Los AngelesCrenshaw LRTSurface/Subway14008.5165180009150
PortlandMilwaukie LRTSurface14907.3204228008952
San JoseSilicon Valley BART MetroSurface/Subway256310.2251467005381
HonoluluHCT MetroElevated534820.12661160002294
WashingtonDulles MetroSurface/Subway314211.7269857003134
Los AngelesWestside SubwaySubway53409593780007607
SeattleUniversity Link LRTSubway19483.16284020015632
Los AngelesRegional Connector LRTSubway13672684900007594
San FranciscoCentral LRTSubway15781.79283510026445
New YorkEast Side Access CRSubway73863.5211016730012614
New York2 Ave SubwaySubway48872.321252130009976

* In essence, this means reducing the degree of provisions for independent rights-of-way for each project is the best way to save costs. But those same reserved rights-of-way are the best ways to keep transit reliable and fast; the effort to reconcile this problem is the raison-d-être of BRT. Data for charts and table above from either agency websites or Federal Transit Administration 2012 New Starts Report.

52 replies on “How significant an opportunity for reducing U.S. construction costs?”

I hearken back to the discussion of the Chunnel construction a few decades ago. The raw tunnel was bored with 11 machines, of which half were scrapped in place. And the net result was two train tunnels and a third service tunnel, too small for trains. But the decision was made to build it quickly, within a political time window. Unfortunately, the raw tunnels cost 8 billion.

An alternative proposed at the time was to use one machine and drill all three tunnels (the service tunnel being the same size) in a serial fashion. Naturally, it would have taken six times longer, but it would have cost only a billion. (of course only an estimate – a realistic number would be 1.5 billion).

If we had the ability and wherewithal to simply bore miles of tunnels with one machine per city (possibly two, depending on rock and dirt considerations) and later figure out where to put stations, how long a platform, etc., it might be both cost effective, as well as put a few more people to work.

For San Francisco, bore down Geary, along 19th Avenue, out to North Beach?

Note that the R^2 for the subway regression says that “subways mostly cost whatever they cost”. In other words, for the subways, there is no evidence presented for the thesis ~ it might well be project-specific costs dominating.

Its only the surface rail projects where the R^2 is strong enough to support the thesis.

Its interesting that a R^2 above .7 is obtained with the intercept constrained to 0 ~ it would be worthwhile for that set to regress the general y=b+ax form and test whether the fixed cost component b is significantly different from 0.

In any event, that suggests that for a surface railway, the target for cost saving is identifying an efficient alignment, and developing a cost-efficient corridor construction process.

One potential cost factor other than direct construction costs that would be difficult to single out in this way is regulatory requirements with respect to alignment approval, since that would be expected to increase in proportion to the length of the alignment.

It would be a useful exercise to focus in on the surface rail projects and go through their costs in terms of project costs, construction overheads, construction payroll costs, and other construction costs and do independent linear regressions of each against corridor distance.

Interesting point. I constrained the regressions to 0 because, well, a line that is 0 miles should cost $0. Of course, that’s not necessarily true, since there may be costs that come with a project of any size, etc.

Here are the equations and the R^2 for the unconstrained regressions:

Subway: y=418.05x+2246; R^2=0.21
Surface/Subway: y=547.46x-3179 (not explainable, since this would make a shorter project free); R^2=0.98
Surface: y=69.33x+57.4; R^2=0.74

And I agree, it would be a useful exercise to evaluate projects one-by-one to look for differences; will probably do as much in the near future.

I wasn’t saying looking at projects one by one ~ I was saying break down the distinct cost categories and look for patterns in the surface subset which seems to substantial enough to yield useful info.

The combined surface/subway would need a proxy variable for subway, the fact that the subway projects are on average shorter is what generates the negative intercept.

We basically need to halt construction until we figure out how to create a system that consistently gets stuff built at the lowest prices they can muster in Europe. Moving forward with projects despite the broken system just guarantees they’ll all fail by any objective cost/benefit analysis.

That said, the discrepancy is probably worse than we think. The long-term dollar/euro exchange parity is probably somewhere around $1.20 or even less if you value things in purchasing power parity, which is the relevant measure when comparing construction costs, so the French conversion that you did in this story overstates the real costs the French have by nearly twenty percent. $30 million a mile is probably the real number there.

For the projects where the benefits justify the costs, there’s no guarantee that indefinitely postponing the receipt of the net benefits in hopes of a bigger net benefit down the track means an increase in total net benefit.

And of course, halting construction of new projects until we create a lower cost system for building rail corridors is mostly a recipe for abandoning rail projects altogether, since a major part of creating a system for lower cost construction is for a variety of projects to be pursued trying a variety of approaches to cutting costs and sharing what works.

I’m glad you pointed out that the high cost of the Portland-Milwaukie Light Rail is partly due to the bridge, but it is also important to make clear that the bridge will be multi-modal. It will carry the Light Rail line, a streetcar line, a large number of bus lines, bikes, and pedestrians. Notably, it will not carry any cars. This gives a whole bunch of transit lines a congestion-free way to cross the river, plus another link for bikes and pedestrians. I really think given all the other future uses on the bridge, it is disingenuous for people to include the entire cost of the bridge in the cost for the light rail line. If has to be included, figure out the percentage of total transit service hours on the bridge that will come from light-rail, and adjust the cost accordingly.

The Portland-Milwaukie Light Rail line is still more expensive than any other rail line when you exclude the exclusive bridge. Minus the bridge cost of $190 Million from the over $1.2 Billion line cost, the 7 plus mile cost comes in around $144 Million per mile. Almost 3 1/2 times what Virginia’s line is costing.

The Virginia line is on an existing rail right-of-way which they took over. The Portland-Milwaukie line isn’t. So, no comparison.

Per mile costs for entire projects are MEANINGLESS and nobody should ever use them. They only make sense when truly comparing like-for-like (surface rail on greenfield with particular soils and grades vs. surface rail on greenfield with same soils and grades, bored tunnel in a particular geology vs. bored tunnel in the same geology, tunnel under road with particular utilities vs. tunnel under road with same utilities of same age, etc.).

I’ve seen the people building the light rail line in Norfolk and they did shotty work building in that you would see them lay down a 500 foot section of cement imbedded track in the street and two weeks later they would rip out the same section of track do to it being built the wrong way and they would rip it out and rebuild it again. Personally they should have fired the cosntruction group building the light rail line and higher someone else who has worked on other major light rail projects.

ESA is not 3.5 miles. The full length of the new route is about 3.5 miles, but most of it, including the underwater part, was built in the 1960s. The length of the new construction is about 1.25 miles.

Mile and quarter in Manhattan. There’s a lot of construction in Queens and the Bronx associated with it.

The Queens construction is a short tunnel portal connection. The other construction is not underground, i.e. is not the cost-buster. (Well, the 1.25 miles of tunnel aren’t the cost-buster, either; the cavern is.)

The Queens construction is much more expensive than you think. Look up what they’re actually doing: they’re running it under and the Queens Boulevard line (underpinning!) and then running one of the tracks under the railyard (underpinning, switches, working in an active railyard)…. pretty much every possible thing which would make construction more expensive. On top of that, Queens isn’t really above sea level, so there’s drainage work.

“I wouldn’t have designed it this way”, as the saying goes. The work in the area of Harold Interlocking has involved a lot of churn and redundant grade separation construction.

Agreed that the cavern station is the reallllly expensive part, of course. Even with the extra underpinning costs it would have made more sense to connect to the existing Grand Central, but Metro-North and LIRR don’t want to cooperate with each other.

Don’t forget that a lot of the project cost quoted in the table includes upgrading basic street infrastructure.

Both the funded Central LRT in the Twin Cities and the planned Northeast LRT in Charlotte are great examples. Each runs largely at-surface and semi-exclusive of traffic with mostly at-grade, signalized intersections. But since each requires significant re-construction of the overall street, you have higher cost per mile for non-transit project elements, such as bike lanes and new sidewalks. Comparatively, Norfolk’s project, outside of Downtown, largely runs on railroad right-of-way, where the project envelope is largely just the ballasted tracks, OCS poles, and substations.

Since riders are often more directly served along streets, instead of railroads, I still favor at-surface, street-running routes for LRT projects. It’s just a shame these projects can’t use less intrusive rapid-streetcar design elements for lower capital costs.

Since riders are often more directly served along streets, instead of railroads,

Very often “downtown” is where the railroad station used to be. So using the railroad right of way will bring you right into the center of downtown – give or take a block or two.

Adirondacker- That’s fine if you want a commuter rail line with lower ridership, albeit at lower capital cost. But LRT should ideally be part of a larger system, serving polycentric destinations for fuller trains in all directions throughout the day.

My examples link multiple downtowns (Twin Cities Central LRT) and/or universities (both Twin Cities and Charlotte NE LRT) with their Downtowns, so that the patronage will be more than peak-period customers headed to the CBD.

Yeah, and even if train stations are in the “downtown” of their suburbs, you can get locals who want to keep downtown quaint blocking any efforts at job or residential density. You see this in Silicon Valley, where Caltrain serves a perfect corridor of old downtowns, but all the new office and residential development happens out by 101 anyway.

Sometimes, in addition, the railroad ROWs leading downtown have been severed by thoughtless construction later on. I can think of multiple examples in Minneapolis (St. Paul, in contrast, still has its lines running downtown). The situation is worse in Syracuse, NY, where the expressway construction destroyed the elevated downtown ROW for the train route.

There are lots of exceptions to the rule that the railroad ROW goes downtown. :-P

Then there are a few OLD cities (Baltimore, Philadelphia) where attempts to locate the railroad station properly downtown were thwarted by NIMBYs back in the 1840s, so the lines skirt downtown. (Philly basically fixed this eventually, Baltimore expanded past the railroad line.) NIMBYs are eternal.

I think their is a lost opportunity in your thoughts, It seems that Transit agencies and the politicians that support them really don’t do as good of job of sounding the overall benefits for the rest of the infrasctructure that is impacted and more times then not improved or pursue changings that might bring that extra bike connection, or replace that aging or re-aligning existing utilities that might serve the community in the long term let alone those stree and sidewalk rebuilds.

Clearly the conclusion is to build passenger rail projects at the surface, in exclusive ROW, and to sever all the roads crossing the rail corridor. Then the road budget will be forced to pay for the reconnection of whatever roads are considered necessary, and the budgeted cost for the rail project will look much smaller.

Only half kidding. There is a nasty habit of budgeting road-rail intersection improvements and changes to the *rail* budget rather than to the road budget. Really they ought to be 50-50.

The 63rd street tunnel is deep under Manhattan. The interlockings for Grand Central start under 58th street. Go ahead get the trains up from tunnel level. Do it around the traffic from Metro North tranis.
Every proposal I’ve seen – they go back to the late 60s – has the LIRR in a separate terminal.

You responded to the wrong comment of mine, but anyway, I have seen MTA engineering diagrams showing a link from the 63rd St tunnel tracks into the lower level of Grand Central; so I guess it’s feasible. Though they always include an additional deep terminal, granted. Wonder what happened to the proposal.

Isn’t some of the reason for the separate tunnel not having to change electrical systems on one system or the other (LIRR vs Metro North)? I could also see adding a link being problematic since you’d be adding switches and capacity/complexity to an already complex station. I’d love to see the proposals for it too.

At the station itself, there are so many tracks it shouldn’t ever be a problem. But even on the approach tracks, it’s possible to have one third rail on each side, one for Metro-North and one for the LIRR.

Cars have shoes on both sides. First one that runs through will rip shoes off, damage the third rail or both.

There are things called “flip-shoes” which are a bit annoying but not that hard to retrofit to the trains, which allow operation on both types of third rail.

That said, really the three different electrical systems is an embarassment. Switch it all to overhead I say.

One item about the The Tide that I don’t see mentioned here that will add value to the LRT in several years and that is the Tide Harbor Park station will be adjacent to an Amtrak station stop in 2013.

Virginia is spending $87 million (of state money no less) to extend Amtrak Northeast Regionals to Norfolk, south of the James River. The plan is to start with one daily train and then eventually expand it to 3 daily trains, all of which will run to DC and NYC. Depending on which NE Regionals are extended south of the Richmond Staples Mill Road Station, could get 1 seat service to Boston from Norfolk. The 2 daily trains to Newport News north of the river will remain on the schedule.

The Amtrak Norfolk service will bypass the route through Richmond to the Main Street station because of the very slow tracks on that route. There is a Virginia DRPT webpage with a short summary of the project status:

The Amtrak connection may not boost daily ridership numbers all that much, but it should boost the profile of the Tide LRT in Norfolk by providing a intercity train connection option to the NEC.

The new Amtrak trains in Vriginia pound for pound are fairly good in terms of cost in that the Republicans in Vriginia haven’t gone nuts to cut or at the least don’t mind extending it.

It also helps when you got a large commuter base to work off and the US Capital as your end point for that commuter base as well as connecting to the NEC.

In the same breath, I think their is conservative support for such intercity service when you can tie together some major urban areas and sometimes a state capital/or large metro area that might not be in the state (In Virginia’s case, DC). Thinking of Michigan’s new republican Gov supporting his states HSR grant (hopefully he will push Amtrak and NS to settle on their issues to move things forward), Wisconsion’s Governor willing to support Milwaukee to Chicago improvements (Wishful thinking, but it would be nice to see another round of grants to see this happen), and Missouri quietly supports the River Runner going between KC and St Louis (Another wish for more grants).

I also wanted to mention Indiana and Ohio. I think their will be some focus from those states if the momentum could keep going in the Midwest, specifically connecting Indy to Chicago as well as Cleveland/Toledo to Chicago. Or another way to look at it, comparable to connecting Norfolk with DC area/NEC that is competitive with air travel in terms of time and reduces congestion on busy freeways.

Another issue, for which St. Louis metrolink cross county expansion is a classic example, is that a lot of the proposals put forth on the ballot box or argued for priority have budget numbers with no real engineering or putting the pencil to paper for everything from ROW acquisition to hard quotes on materials, etc. In St. Louis, the budgeted number was based on at grade pricing. The changes were significant as they drastically increased the amount of below grade trackage, significant increase in utility work, and more elevated line. Unfortunately, the media played it as a cost overrun from the outset.

In other words, we created a system were very little engineering is done and then and only when revenue source or tax support is approved do we get into nuts and bolts of things on paper. I think the stimulus offered planning grants, would like to see more of it incorporated into transit funding.

True, betterments demanded by the (immediate) community aren’t “cost overruns.” However, such project design changes or non-transit elements are non-essentials.

Hence, the Feds should require higher local match of those projects spending a larger portion of their project budget on such non-essentials. The Cross County extension, which was entirely locally funded, wouldn’t even likely scored a good enough CE rating to be a New Starts project.

What a clever spammer you are, copying someone else’s comment earlier in the thread so it looks like you are a valid commenter. It seems that the depths that spammers will sink to are far lower than any envisioned Manhattan termini…

Another way they could cut costs building these giant train stations for the high speed rail stations is if they based the new stations of a 1930’s art deco shape vs. this ultra modern crap which requres a lot of useless angels and sections of glass and metal pointing out every with way. Also the 1930’s buildings have been able to use a lot less repairs then these ultra modern train stations that have a lot of other useless things about them that add to their costs.

It’s cheaper to build a sqaure shapped building then it is to build one with all these crazy angels and shapes in them.

Stations are a really small portion of HSR construction costs. The only stations that cost a lot are greenfield underground ones, such as Transbay Terminal, and with those, it’s not the above-ground architecture that matters.

For small-scale operations, station costs could be significant, if the stations look like full-service old-time depots. For regional and small-scale intercity service, platforms with level boarding, a timetable, shelter, a PA system connected to the dispatcher, and a few parking spots should be enough.

The US has strange conceptions of railroad stations based on old-time depots. I remember people being surprised at the idea that any enclosed waiting room should be directly on the platform (which is standard in most of the world) because That’s Not How Old US Stations Did It.

I’ve seen some miserably deteriorated 20’s/30’s buildings from all styles including deco. A lot of the worse of it is already gone, torn down 20, 30, 40 or more years ago – it had a short lifespan, being commercial (hence designed for a short lifespan) and going in and out of fashion or not meeting more modern needs. Plus they often didn’t get the detailing right and the buildings, the less-well-detailed, were on their last legs.

A well built and well detailed building of any style will stand the test of time (if not fashion).

Forgive me if this is a silly question, but why can’t we just break down the costs of a few major American projects and compare them to the broken down costs of their European counterparts. When I mean this, I mean going and looking at the exact costs of tunneling, trackwork, embankments, labor costs,etc and looking at how much they deviate, along with where and under what conditions. Otherwise, (as is apparent from the closing paragraph and the majority of comments) we really can’t say anything more than what we already know.

I don’t understand why there is a perception that cost-control problems are worse in the USA than in Europe. The UK (which is in Europe) has major problems with keeping transport projects on time and within budget. The Edinburgh tram (streetcar) scheme has been such an unmitigated and widely publicised financial disaster that the prospect of any other UK city building a new tram or light rail system (as distinct from extending an existing system) is now virtually nil.

Well, the Edinburgh case involves a giant spat between the city and the contractor, who are now suing each other. That *never* goes well.

That didn’t happen in any of the previous UK tram schemes, so if people are using THAT as a reason not to build light rail… that’s kind of stupid. If they’re using it as a reason to avoid either the contractor or the city government, that would make sense.

It would be nice to the cost/mile and cost/rider-mile for Vancouver, separating the older elevated lines from the newer subway line and perhaps adjusting for inflation since the older lines were built some time ago. I expect costs might compare favorably, especially per rider-mile.

Two huge costs are utility relocations and park & ride facilities. Here in Phoenix the park & rides cost about $7 Million for commercial land & construction costs. The utility work can get extensive since you may have to move a water line to the side of the street for 3 miles, then repeat for sewer, storm drains, gas lines, etc. Or encase them in stronger sleeves do to the light rail weight of 100,000+ pounds per car plus passenger loads.

Finally, I would also look to see how much was spent to make the stations look pretty vs functional usage. Art work to dress up the stations runs about 2% of total station project costs for the Phoenix light rail from what I’ve seen.

How is it that the USA built an enourmous passenger rail network, including streetcars and interurbans, between the Civil War and World War 2?

At today’s per mile cost, we would be utterly bankrupted today if we tried to replicate the network that we scrapped only yesterday [historically speaking].

While this may seem to point to a mistake we made by not continuously updating a national passenger rail network, as Europe did, that is not the point here.

I’m simply making an empirical inquiry: how could a much less wealthy nation — the US of that time period — afford such a large investment? What’s the difference between then and now? Labor costs? Safety? Technical specifications, what?

Leave a Reply