13 November 2015

The U-Shaped Grade Separation

While some cities and towns on the peninsula are still holding out for trenches or tunnels to bury the railroad tracks out of sight, the astronomical cost and difficulty of constructing such structures below the water table in seismically unstable soils makes it likely that above-ground solutions will ultimately prevail, anywhere rail traffic needs to be separated from road traffic.  An attractive above-ground solution is the U-shaped grade separation.

What is a U-shaped grade separation?

U-shape bridge cross section, showing the benefits of
reduced track elevation
A U-shaped grade separation is a type of railroad bridge used to elevate the tracks above road traffic with as few community impacts as possible; there are no property takes and all road turning movements are preserved.  The bridge structure consists of sections made from two pre-stressed concrete side beams, forming the two sides of a U shape, connected by a flat slab forming the bottom of the U, on which the tracks are laid.  The side beams bear the bending loads from the weight of the bridge and the trains that it carries.  This is not a typical railroad bridge design; it is a specialized configuration used to quickly and efficiently build elevated urban metros in cities where these systems are being built from scratch in a densely built environment.  The concept is further explained in a paper and a patent.

While the peninsula rail corridor is not a new metro system, these U-shaped structures could still prove useful in a major push to grade-separate the 40 grade crossings that remain, enabling higher speeds and more train traffic while relieving road congestion and improving east-west access across the tracks.

What are the advantages of U-shaped grade separations?

U-shaped grade separations combine several attractive features that make them ideally suited for developed areas along the peninsula rail corridor, and certainly much better than the massive hollow core concrete box girder bridges considered standard issue by the HSR project as shown in the graphical comparison above.
  • Lower track elevation.  The U shape minimizes the depth of the structure (measured from the underside of the bridge span to the top of the rails) to 3 feet or less.  This allows the standard 16-foot road clearance to be provided by raising the tracks just 19 feet above the road surface, about 8 feet less than the large elevated concrete box-girder viaducts that were proposed during the 2010 Analysis of Alternatives for peninsula HSR.  The rails are lowered thanks to the U shape, which places the structural support of the bridge to the sides, rather than under the trains.
     
  • Lower visual impacts.  When the tracks don't need to rise as much, the rail approaches to a grade separation become correspondingly shorter and less obtrusive, impacting fewer views. The structures above rail level, such as overhead electrification poles, are also lowered.  This reduces the so-called "Berlin Wall" effect of a grade separation structure.
     
  • Lower train noise.  The side beams function as natural sound walls, trapping rail noise before it has a chance to escape into adjacent neighborhoods.  They are especially effective because they are thick and quite close to the train.  This obviates the need to add sound walls on top of the bridge, making the finished structure less visually obtrusive.
     
  • Better earthquake resistance.  The lower profile of the bridge structure reduces bending moments applied to the piers and foundations, whether by earthquake forces or train braking and acceleration or wind loads.  This makes the bridge piers less massive and integrates them better into the built environment.
     
  • Better station integration.  Where stations must be located on an elevated section, structures are simplified thanks to the lower profile of the track, which reduces the reach of stairs, ramps, escalators or elevators, making for a more passenger-friendly environment.  The side beams of a U-shaped viaduct have their top flange at the same height as the train floor and form the actual platform interface, 50 inches above the rail and 72 inches from the track center line, allowing the U-shaped structure to continue uninterrupted through the station.
     
  • Better safety in case of derailment.  The side beams are close to the train.  In case of a derailment, train cars will be guided by the structure and will not topple off the bridge.  This feature is known as "derailment containment."
     
  • Lower construction cost.  U-shaped elements can be prefabricated off-site and assembled with minimal disruption compared to traditional cast-in-place construction methods.  Using standardized elements throughout the corridor, in dozens of locations, provides economies of scale.  The decreased profile changes for both rail and road (whether the U-shaped bridge is elevated or at-grade with the road sunk underneath) require less excavation or fill.
The U-shaped design can minimize property takes, preserve turning movements for cars and trucks, cost much less to build than below-grade solutions, and tread more lightly through built-up neighborhoods than a conventional (box beam) viaduct or split-grade separation.  U-shaped bridges are ideal for grade separation in dense areas like the peninsula.

38 comments:

  1. Setting aside the fact that we won't be quad tracked for a while, can this also support four tracks, or would you need two U's for that?

    ReplyDelete
    Replies
    1. Two U's side by side, or a triple-beam arrangement (dare we say "double-U"). With the top flange of the beam 6 feet from track center, a 3-foot-wide flange fits in the standard 15-foot track spacing.

      Delete
    2. The Caltrain line still has some triple-beam bridges with one track between beams. They're being replaced and Caltrain won't see triple-beams again until HSR makes it quad track

      Delete
  2. The concrete version in the drawing looks elegant, but is really nothing more than a variation of the traditional "half-through" bridge configuration. Given that Caltrain is currently in the midst of installing four half-through steel girder bridges in San Mateo, they clearly already understand the benefits in height constrained areas. Maybe they'll consider your fancy concrete U-shaped bridges for Palo Alto.

    ReplyDelete
    Replies
    1. As I said in my reply to Shawn Wilsher's comment, I said that the old San Mateo bridges are triple-beam, and Caltrain will not see triple-beams again until it is quad tracked for HSR.

      Delete
  3. Caltrain San Mateo bridges typical vertical:

    Rail + tie plate: 8 inches (0.203m)
    Concrete tie: 8 inches (0.203m)
    Ballast: 8 5/8 inches ± 1 inch (0.194m – 0.244m)
    HMAC: 3 inches (0.076m)
    Waterproofing: 1/4 inch (0.006m)
    Deck plate: 5/8 inch (0.016m)
    Floor beam: 23 inches (0.584m)
    Gap: 1/2 inch (0.013m)
    Bottom flange: 2 inches (0.051m)

    Total: 4 feet 6 inches (1.372m)

    ReplyDelete
    Replies
    1. Your point seems to be that the U-shaped bridge structure shown in Clem's speculative drawing saves 18 inches in construction height compared to a real-world steel plate girder half-through bridges in use in the Caltrain corridor. My only point was that half-through bridge structures, whether in concrete or steel, are well-known and nothing new. Clem hasn't mentioned how he decided on the deck depth for his drawing. The paper only mentions "between 0.22 m and 0.35 m" for "LRT and MRT viaducts", while the patent provides no numbers at all.

      Since you seem to have all of the drawings at hand, what bridge configuration is used in the overcrossing adjacent to the new San Bruno station?

      Delete
    2. Marc, I had no "point". I was just supplying some data. (And sorry for mis-nested comment; blogspot.com's terminal bit-rot continues. I'm hacking raw HTML in two different browsers to try to post a simple damned comment.)

      I didn't dig deeply (there were thousands of drawings and several addenda), but it looks like the San Mateo and San Mateo Avenue underpasses in San Bruno were specified with nominal 66.625 inches (1.692m) from top of rail to base of structural steel. (28 inches 0.711m from TOR to deck plate, 0.625 inch deck plate, 38 inch 0.965m deep "W36X361" 61 foot 18.59m long longitudinal beams for each span spaced at ~0.784m.) Angus Avenue is deeper at 77 inches 1.96m (= 28 inches to deck plate, 0.75 inches deck plate, 3.5 inches top flange, 41.25 inches web, 3.5 inches bottom flange, beams spaced 0.762m) for its single 84 foot 25.6m span. There are no big side "U shaped" side girders for these bridges, but rather a dense array of identical longitudinal beams under the deck.

      In contrast, the raised San Mateo bridges have spans of 14.770m (Poplar), 14.564m (Santa Inez), 14.605m (Monte Diablo), 14.710 (Tilton) and are constructed ("U shaped") with honking 5 foot 8 inch 1.727m tall side through girders that bear perpendicular 0.584m tall "W21X201" floor beams spaced at 0.781m that in turn support the deck.

      Delete
    3. I apologize for misinterpreting your comment. Caltrain can reduce the construction height of overcrossings when they want to, and this makes one question whether the San Bruno berm was constructed higher than perhaps would have been necessary with an alternative bridge configuration. I'm no engineer, but I do suspect the diagram Clem provided underestimates the distance from TOR to the underside of the U-shape deck, as it would likely need to be deeper than an equivalent steel structure.

      Delete
    4. Keep in mind we're comparing different structure types. The deck can be thinner if the bending loads are transferred out to the side girders... The deck then consists of transverse elements, not unlike slats under a mattress. That is unlike the San Bruno and San Mateo examples, which have beefy (deep) longitudinal beams under the track.

      Delete
    5. Clem, the paper you linked mentions the need for some combination of pretensioned lateral beams, precast segments, and/or poured in place slab. None of this is going be thinner then an equivalent combination of steel deck plates and lateral beams, as used in the San Mateo half-through bridges. Please do a bit of research on half-through bridges, and you'll understand the difference. The box girder bridges used elsewhere on Caltrain require much deeper longitudinal beams underneath the track, half-through bridges move the girders to sides at the deck level and above.

      Delete
    6. I'll also mention that those elegantly thin cross sections illustrated in the paper are for LRT systems, fully loaded freight trains will require much thicker decks and side beams.

      Delete
    7. How would ballastless track come into the calculation?

      Delete
    8. This site, chronicling reconstruction of a stretch of rail in Berlin, has a good example of a thin steel bridge. Mainline passenger services use the route, as well as the S-Bahn.

      http://www.ostkreuzblog.de/baufortschritt/baustellenrundgang-ende-september-2015/

      The example is in photos 3,4, and 5 as you go down the page, with 5 showing the depth the best. I think the arch span is a great solution for longer spans. Who doesn't love an arch?

      Delete
    9. To add some more to Richard's details about Caltrain bridges. Jerrold Ave was replaced a few years ago with a 134' steel, double track, ballast deck, through plate girder span. This is, in its simplest form, is the steel version of a "U" shape bridge. The girders are about 13' 6" high. Track centers on the bridge are 15' 0" and the center to center distance of the girders is 39' 6". Top of rail to low steel is about 5' 6".

      Depth of the structure is a function of the span length and the distance between the girders. Jerrold uses 30" beams for the floor system.

      Similarly, the new span over Angus Ave at San Bruno is a multi beam system and has very deep members driven by the length of the span (84') and the skew required to meet the alignment of the street.

      Concrete still has to obey these rules, so length of opening and width of structure will be key items in determining how deep it is.

      Delete
  4. As a random observer I'm curious: this appears to be a solid list of advantages with no disadvantages. Is there a disadvantage (perceived or real) that would ever lead to selecting another option?

    ReplyDelete
    Replies
    1. Design choices, engineering constraints, cost, lead-time, availability, stakeholder preference, etc., take your pick. A simple thing like whether you can get a suitable crane to a particular site can require using an entirely different construction approach.

      Delete
    2. The perceived disadvantage will be that this is an elevated structure (keyword Berlin Wall). I'm trying to point out that elevated structures need not be overbearing to their surroundings if designed carefully.

      This contrasts with the paint-by-numbers design that was proposed back in 2010, where massive freeway-like viaducts were to be rammed through many cities.

      Delete
    3. Clem, have you forgotten that you titled this post "The U-Shaped Grade Separation", and that it talks only about railroad bridges above roads? For extended viaducts, I agree that this approach is aesthetically more pleasing than either the BART-style or (for that matter) NYC-style elevated structures. Grade separation bridges between berm segments are a very different matter.

      Delete
    4. @Marc. The berm segments will also be lower. Besides which to make the berms more aesthetically pleasing, add more U-Shaped bridges along with landscaping.

      Delete
    5. One additional disadvantage may be that spans can not be as long. But this is not really such an issue if it is for bridging roads.

      Correction: The Letzigrabenbrücke in Zürich (longest rail bridge in Switzerland with 1156 m) is u-shaped, single track, and the longest span is 60 m. A short technical description can be found under http://www.bp-ing.ch/files/referenz_5665_durchmesserlinie_letzigraben.pdf (PDF document, in German language). Height over all is between 3.3 and 3.6 m with a width of 10 m.

      So, we can say, this bridge type is pretty much state of the art… 

      Delete
    6. One thing I forgot to mention about that bridge; it has to cross other tracks, but the line has to be down to the ground quickly, because it has to go under an existing road bridge. That means that there should be a minimum of understructure under the tracks, in order to keep the grades under control (and if I am not too mistaken, it still has 3 or 3.5%).

      The bridge also uses ballastless track, which is a bit noisier, but needs less maintenance. The u-shaped cross section acts as noise barrier (and if it would get too bad, there are possibilities to add noise attenuation panels).

      Delete
    7. Ballastless track also lowers the profile (ballast requires extra depth) which is the other reason it's used besides lower maintenance.

      Delete
  5. And where they have enough room to also lower the roadway, then even lower profile! My recommendation to the engineers: Be creative!

    ReplyDelete
  6. The outer beams also obviate guard rails. (File under "Lower Visual Impacts".) A (light rail) example is in Sacramento along Folsom Blvd. The Watt Ave overcrossing was built circa 2009 and has three beams supporting two tracks. The overcrossing at Sunrise Blvd and the undercrossing at Power Inn Rd are older, more conventional beam designs.

    Watt Ave:
    https://goo.gl/maps/Hr51LNuteTs

    Sunrise Blvd:
    https://goo.gl/maps/zkBap3LfzX62

    ReplyDelete
    Replies
    1. When I saw both of those grade sepatations in real life, it reminded me of Hamilton Avenue in Cambell, CA where there is a grade separation for light rail right next to a freight branch line. Here it is https://www.google.com/maps/@37.2943673,-121.9352806,3a,75y,264.86h,80.22t/data=!3m6!1e1!3m4!1sMivSpfKGX4Dj6Gm3pD4l1w!2e0!7i13312!8i6656

      Delete
  7. Alright, I just stumbled onto this blog, and have been reading it for hours. What a wealth of information, totally fascinating. Too bad it is going to take 20+ years to get things up to snuff even with the best plans laid out here in extensive detail.

    Here's what I now wonder: At some point, wouldn't it be better to just replace the passenger rail lines with lightweight single lane roadways, and just buy 1000+ Teslas with auto-drive? Super fine time granularity, high average density, minimal staffing, new "rolling stock" can be bought within months, no track grinding, etc.

    ReplyDelete
    Replies
    1. I'm not quite sure why everyone seems to think that self-driving cars will solve all of our problems. Safety dictates that there must be at least a safe stopping distance between vehicles (something which regular drivers routinely ignore) - as Clem mentions, it still takes multiple lanes to get the capacity of even a moderately used rail line.

      And the energy efficiency of electric vehicles is never going to come close to that of electrified rail. That's just a fact of rubber tires vs steel wheels on steel rails.

      Delete
    2. Well, in a massively automated system, with both distributed and central control, the stopping distance wouldn't be *that* big an issue, I don't think. If every car knows the planned speed trajectory of the car ahead of it and behind it.

      And it would be targeting just a single lane system, so you save all the complexity of lane management, but at the cost of serializing the entire route.

      Really, the basis of the idea is that it is doable, since a boatload of cars could just be ordered from Tesla. Whereas, as I am reading from this blog, all the issues of the weight of trains, track gauge, tech compatibility and so one are really big problems. Whereas, we are pretty good at building simple roads.

      Delete
    3. Automobiles do unplanned things. So do other modes.

      Delete
    4. Capacity still doesn't scale. That's a fatal flaw.

      Delete
    5. Safe stopping distance limits capacity regardless of whether a system is automated or not. Even the best systems fail occasionally, and such errors have to be recoverable without loss of life.

      Delete
    6. (That, by the way, is the fatal flaw of the Hyperloop... but I digress)

      Delete
  8. Cars (self-driving or otherwise) don't scale to the massive capacity of rail. At 5 tph today Caltrain has a demonstrated peak throughput of about 4000 people per hour per direction or ~3 freeway lanes. Caltrain is slow and infrequent by world standards. Train frequency and length will more than double this throughput in the next decade, primarily using the existing tracks. Your Tesla idea has the same flaw as the Hyperloop: insufficient throughput. That's why the trusty old "19th century technology" of rail transit isn't going anywhere, no matter what the self-driving car enthusiasts may say.

    Oh, and Teslas lug around their big heavy battery, which is horribly inefficient. EMUs use a single pantograph to suck six megawatts off the electric grid, with no dead weight on board the vehicles.

    Thanks for reading.

    ReplyDelete
    Replies
    1. Well, the Teslas do have the battery weight, but I wonder how the weight/person works out in the CalTrain or BART cases. Those are big machines after all. It looks like CalTrain is ~110Klbs per car and ~260Klbs per locomotive. Info I can find seems to indicate that capacity is about 700 people for the whole assembly, at peak times. So that's ~1300 lbs/person,or 1150 lb/p going by 800 riders per train. A 4830 lbs Model S with 4 people gives ~1200 lbs/person. A custom 6-person interior could drop that to 800 lbs/p.

      Delete
    2. Vehicle weight per passenger is not a meaningful efficiency metric, particularly when comparing rubber tires to steel wheels and rails.

      Delete
  9. Is 90cm between bottom of bridge and top of rail really the best that can be done?

    ReplyDelete
  10. Does the U shape require two edge columns instead of one?

    ReplyDelete