15 December 2020

Redwood City Grade Seps: We Must Do Better

The first preliminary engineering plans for the downtown Redwood City grade separations are out. This isn't your average grade separation: it underpins the most important new piece of rail corridor infrastructure that will enable future Caltrain service to be much better than it is today. Redwood City will become the main overtaking location for Caltrain, with a cross-platform transfer between local and express trains at a new four-track station. We had previously looked at Redwood City issues and laid out the design values that will make Redwood City a high-functioning station.

There are four grade separation alternatives on offer, as described in the overview video:

  1. Fully elevated all the way from Whipple to Highway 84
  2. Partially elevated, in two phases, with Jefferson rebuilt in phase 1
  3. Partially elevated, in two phases, with Jefferson rebuilt in phase 2
  4. Partially elevated at Whipple, with everything else staying at grade

If you're going to be driving a car, the good news is that the designers have done a fantastic job with the car infrastructure, with an A+ on road and intersection design. Unfortunately, this is a grade separation first and a train station second, almost as an afterthought. It should be a train station first. To understand why, let's look 20 years ahead.

Better Service for Many More People

In 2040, which sounds futuristic but is relatively soon, rail service in Redwood City is far better than was ever imagined in the depths of the pandemic of 2020. Here's what is happening:

  • Frequent peninsula rail service. As planned in Caltrain's service vision, electric trains serve a vibrant downtown Redwood City with all day half-hourly service and up to eight trains per hour per direction during the morning and evening peaks, equivalent to adding 5 new lanes to Highway 101. Except it's centrally located, it's quiet, it's emission-free, and it's faster than the hordes of electric autonomous cars gridlocked on 101. At the new elevated four track station, every 15 minutes, in each direction, a local and express train stop side by side on opposite edges of the same station platform. Express passengers who are going to minor destinations transfer to the local, and local passengers going to major destinations transfer to the express.
  • A new station at North Fair Oaks. Following the closure of Atherton station, eliminated for lack of ridership in 2020, a new Fair Oaks station has been created near Fifth Avenue and the southern end of the four-track segment through Redwood Junction. Perfectly spaced halfway between Menlo Park and downtown Redwood City, this station serves neighborhoods that are among of the most densely populated along the entire peninsula rail corridor. The 2010 census showed more than 15,000 jobs and 35,000 historically disfavored residents within a 1-mile radius, and the area has flourished since then with new rail service. Not only is Fair Oaks a significant source of new ridership and revenue for Caltrain, but it is served with zero additional trip time and train operating cost. We will discuss below how this feat of magic is pulled off.
  • New Dumbarton rail service. Inaugurated in the 2030s and using a new Dumbarton rail bridge to the East Bay and beyond, fast half-hourly rail service is relieving regional congestion and making commutes more pleasant and productive. The Dumbarton rail corridor joins the peninsula rail corridor at Redwood Junction, where it merges without interference to peninsula train traffic to create an efficient transfer with Caltrain at the bustling station in downtown Redwood City.
  • Statewide high speed rail service. While Redwood City was not considered a viable station stop in the late 2010s when the statewide rail network was planned, in 2040 it's a no-brainer to connect to it. Redwood City's focus on growth, connectivity and equity has allowed it to outshine more ossified and backward-looking locations like Palo Alto that once embodied the dynamism and innovation of the region. For this purpose, the platforms at downtown Redwood City are 1300 feet (400 meters) long to accommodate double-length high-speed trains.

Focusing on the track layout, the common thread of these four improvements to rail service is the quadruple tracks through downtown Redwood City, connecting the new four track station to the four track segment at Redwood Junction / Highway 84 built at the turn of the century. These quadruple tracks enable parallel train movements into and out of the Redwood City station, making optimal use of the 80-foot width of the rail right of way. This four track layout brings us back to the downtown grade separation project being planned twenty years before, namely now.

The Fatal Flaw

None of the Redwood City grade separation alternatives allow four tracks all the way through downtown. What's worse, the two-track layouts of all four alternatives are wasteful of the scarce and valuable downtown right of way, making no allowance for adding these critically important station approach tracks later. These tracks are required for two very important operational reasons:

  1. Approach tracks allow parallel and independent train movements into and out of Redwood City from the Dumbarton rail corridor without introducing train path dependencies, and thus time keeping vulnerabilities. A train from San Jose should be able to approach the station in parallel with a train from Union City.
  2. Approach tracks allow efficient overtakes, supporting a better implementation of Caltrain's future service vision. Four tracks allow local trains to serve Fair Oaks instead of wasting six to seven minutes at Redwood City waiting for an express to overtake them. A similar operations concept is possible to the north, if San Carlos station is rebuilt with four tracks.
We mentioned earlier that service to Fair Oaks could serve many people and generate new fare revenue in exchange for zero additional crew labor and equipment cost. How is this magic even possible? Let's compare two operating scenarios:

Scenario A. In Caltrain's service vision, express and local trains overtake each other every 15 minutes at the downtown Redwood City station, which will have four platform tracks to allow convenient cross-platform transfers. Unfortunately, with two-track approach bottlenecks from the north and south, trains must enter and leave Redwood City sequentially. That means the local arrives first, waits three minutes for the express to catch up and arrive behind it, then waits another minute for the express dwell time, then waits another couple of minutes for the express to leave and pull far enough ahead. Only after six or seven minutes, even if everything is running perfectly on time, can the local leave Redwood City. Sitting still at a red signal is operational poison, wasting passenger time (all the more so because the perception of delay time is magnified more than 2x by immobility) and driving up crew labor costs ($/passenger-mile). Meanwhile, there is no rail service for Fair Oaks residents, and a long gap in Caltrain coverage between Redwood City and Menlo Park.

Scenario B. The express overtake occurs on a passing section with one or more stations served only by the local. The deceleration time, dwell time, and acceleration time associated with that extra local station stop allow the express to gain on the local while the local is being useful and providing service, instead of just sitting around for an interminable dwell at Redwood City. In the southbound direction, the local arrives in Redwood City, waits three minutes for the express to arrive, exchanges passengers, and leaves the station at the same time as the express. The two trains run side-by-side until the local stops at Fair Oaks. By the time it's ready to go again a couple of minutes later, the express is long gone and no additional waiting is required. The overtake took the same overall time (six or seven minutes) as Scenario A, but used the time productively to provide local service every 15 minutes at a new station serving 35,000 people within a 1-mile radius, at zero additional crew labor or equipment cost.

Here's what the track layouts look like, with station distances roughly to scale, comparing the current condition with Scenario A (planned grade separation configuration) and Scenario B (with Fair Oaks station):

Bonus points are awarded for leaving sufficient room to run a fifth grade-separated track from the southbound tracks to the Dumbarton corridor, flying over Highway 84, as shown in the last panel. This allows seamless merging of the two corridors, without the two streams of rail traffic fouling each other on the way to and from downtown Redwood City. The right time to plan for this is now, not after we pour concrete in the wrong place, even if the funds aren't yet available and there isn't yet a viable Dumbarton project.

Layout of new Fair Oaks station, roughly to scale
Here's what the new Fair Oaks station might look like, near milepost 27. It's about 100 feet wide in an area where the corridor is only 80 feet wide, so it will require taking nine properties along William Avenue. While taking residential properties in a less advantaged neighborhood of unincorporated San Mateo County is never desirable, this must be weighed with the wider benefits to the 35,000 residents of North Fair Oaks and other neighborhoods within a one-mile radius. A narrow slice (~10 feet) of land would be required to be taken from backyards along the opposite side, to make room for the southbound platform. The site has good pedestrian and bus access via Fifth Avenue, and a new pedestrian underpass would connect Berkshire Avenue across the rail corridor, improving neighborhood connectivity. The new Fair Oaks station is very low hanging fruit that can make Caltrain serve more people more efficiently from day one of the new service vision.

The marginal cost of building the grade separations with the additional tracks now is small relative to doing it later, so why would we delay such an important operational improvement or waste money re-doing it twice?

Not allowing for quadruple approach tracks is the fatal flaw of the downtown Redwood City grade separation project. All the alternatives feature two-track bottlenecks that fail to adequately support Caltrain's service vision and impair the future Dumbarton rail corridor service, which is especially concerning because Caltrain and the San Mateo County Transportation Authority seem to have directly participated in the design.

Here are some other comments on the various alternatives:

The Good

  • The roads and intersections and turn lanes and all the car stuff is perfectly designed for smooth and unimpeded car operations.
  • All alternatives feature the four track station, consistent with Caltrain's future service vision, which calls for Redwood City to be the location where express trains overtake local trains.

The Bad

  • All the bridge structures have porky structure depths, ranging up to 9 and even 17 (!!) feet. Every foot that your bridge deck is thicker is a foot that the entire edifice, including all embankments, will be taller. In this case, each extra foot of structure depth is worth about 1,500 18-wheel dump truck loads of dirt to fill the embankment one foot higher! Thicker structures also push roads to be depressed more deeply into the ground, which exponentially increases excavation and utility relocation costs. Thicker bridges at stations needlessly extend stairs and ramps. For the love of Redwood City, use thinner structures! Previous grade separations, such as the one in San Bruno, have used steel beams to keep the structure depth (from soffit to top of rail) to about 5 1/2 feet for an ~85 foot span. Through bridge designs can be even thinner. Grossly excessive concrete bridge structures as seen here are an unmistakable symptom of not caring about costs and impacts.
  • Enormous extra costs are being incurred at the southern end of the grade separation by not allowing the vertical curve at Highway 84 to begin until north of the overpass and the existing turnout, and then limiting the grade to only 1%. This is lazy paint-by-numbers engineering that blindly applies design standards without regard to their consequences, which in this case push Chestnut Street and others much deeper underground than is necessary. If there ever was a case where an exception to Caltrain design standards was warranted, this is it. Using a 2% grade and starting the vertical curve south of Highway 84, squeezing every inch of available vertical clearance under Highway 84, will save millions. In a modified Alternative 1, Main and Maple may not even need to be sunk at all.
  • The platforms are 900 feet long, and there may not be sufficient clearance left between the station tracks to extend them to 1300 feet (400 meters) later to support high-speed rail service at Redwood City. While this is not in current official plans, it makes sense and it would be a shame to preclude it. There is plenty of room in the site to allow it.
  • The station should be open underneath, not built on a filled embankment. It should allow a future Broadway light rail line to shoot straight through, right under the Caltrain platforms. As proposed, all the alternatives torpedo the city's Broadway street car project.

The Weird

  • The vertical profiles have unusual constraints on vertical curves, seemingly not allowed over or under bridges. That is just silly and grows the embankments unnecessarily taller.

  • Hopkins Avenue should be re-connected across the tracks. It's basically a freebie to improve neighborhood connectivity.
  • All alternatives, not just 2 and 3, should start near Howard Avenue in San Carlos, extending a four track station approach as far north of Whipple as possible. In the future, if the San Carlos station is rebuilt as a four track overtaking station, dwell times at Redwood City can be further shortened than is possible with just the Fair Oaks station, using the same overtaking principle.

  • Pennsylvania Avenue is treated as a city street, not the encroachment on critically important railroad right of way that it actually is. It will be needed anyway for temporary shoofly tracks during construction. The grade separation project should not give away valuable railroad right of way to automobile uses.
  • Alternative 4, at grade, has no clear way to access the station platforms and forms an even more formidable barrier through downtown, seemingly in contradiction with project goals.
The Redwood City grade separation is about much more than just grade separating roads to make traffic flow freely. Like any project on the rail corridor, it needs to be planned and built not only such that future improvements to rail service aren't made more difficult or impractical, but to start putting in the hooks for those improvements now. The marginal costs aren't zero, but they're so much less than fixing it later. Caltrain has a deplorable track record with future-proofing grade separation designs: the San Bruno grade separation and station, just recently completed in 2014, is already officially planned to be partially demolished for the high-speed rail project to straighten out a curve that could have been built correctly in the original design (reference HSR San Francisco - San Jose project DEIR Volume 3, Alternative A, Book A1 plan and profile drawings, sheet 7).
The same kinds of silly mistakes are now starting to be made in Redwood City, with an efficient layout to facilitate express overtakes and Dumbarton service being fumbled by oblivious consultants who excel at the road details and don't seem to appreciate the rail details. When they've got a hammer, the whole world looks like a thumb.


  1. Great work as always Clem.

    RE building on a viaduct instead of retained fill: people have an irrational fear of elevated viaducts. They are actually fantastic, and in many ways aesthetically preferrable to every option short of a tunnel, even including a trench.

    (1) In addition to generally being simpler to construct, and contrary to what some might consider common sense, Higher is actually *better* from an aesthetics and impacts perspective. This implies that the full elevated option is the best on all counts. Here is why: higher viaducts feel more open, let more light beneath, and moves the noise further from ground level. Split grade separations are less desirable; only do these where absolutely necessary - ie, there isn't enough distance to climb high enough.
    (2) If the tracks are high enough, in addition to just putting station facilities under the viaduct, you can put retail spaces under there too. This would be a fantastic fit for a dense downtown like Redwood City. Commonplace overseas, this practice is less common (but not unheard of) in the US: https://www.google.com/maps/@41.96603,-87.6578421,3a,75y,294.13h,82.47t/data=!3m6!1e1!3m4!1s6wsQ9KdQbGVRb91_-IRksg!2e0!7i16384!8i8192 If you're going to do this, the design can't be phoned in. It has to be planned with this in mind. No 17 foot thick bridges.

    1. In my experience the fear of elevated structures is universally rooted in racist and classist attitudes about unhoused persons and "neighborhood character."

    2. If they fill the space under a viaduct with shops, then that pretty much solves both problems anyway?

  2. Clem, this is a fantastic analysis. What can be done to reverse neighborhood fear of the fully elevated model?
    Is there a actual model (as in table model) you know of?
    I would be willing to contribute to one.

    1. Risking a simplistic comparison to a "Berlin Wall", the Stadtbahn across Berlin, built in the 19th Century is beautiful. It is four tracks, has examples of slim over crossings, but is primarily a series of arches. There are markets and restaurants under the arches. This blog is very good at text, but I need to get off my hiney and do a visual tour of what a beautiful railway like Berlin looks like. For starters... https://www.google.com/maps/@52.52214,13.4016903,3a,75y,25.22h,89.6t/data=!3m8!1e1!3m6!1sAF1QipNiG8RJnz9j99tO3Purq8RHymmoHAyNmvZv4HWk!2e10!3e11!6shttps:%2F%2Flh5.googleusercontent.com%2Fp%2FAF1QipNiG8RJnz9j99tO3Purq8RHymmoHAyNmvZv4HWk%3Dw203-h100-k-no-pi-1.3025028-ya304.78525-ro0.9844451-fo100!7i6912!8i3456

    2. I made a video slideshow. It plays by itself. If you need to pause it, pause the movie. Shows some of my favorite rail viaduct examples that would all work well anywhere along Caltrain.


    3. All those examples are excellent; however, they are all European. It is also common in Asia, especially Japan. But Lest we think this is a "Europe thing" or an "Asia thing" that is not possible here, there are also good examples from inside the United States.

      The most recent one I am aware of is Brightline's Miami Central station. It also is a clear demonstration of the principle that when it comes to urban viaducts, *HIGHER IS BETTER*: https://goo.gl/maps/9fdJqUsLzJv3Q5Vk6

      Metra Market in Chicago at Ogilvie Station. The platform area that has shops underneath was built in 1912: https://goo.gl/maps/LdDMgYymziuude3X8

      CTA Wilson Station, also in Chicago, which is an interesting example in that the retail space is apparently not integrated into the viaduct structure, but is instead a separate structure underneath the viaduct, likely with its own roof, rather than relying on the viaduct's waterproofing to keep the interior dry. In this case, the building is old, but the viaduct was recently rebuilt. https://goo.gl/maps/JiPRBAgnWyLnKosx5

      There are likley more than a few examples in other places, such as New York, Philadelphia, or Boston, but I'm not as familiar with those areas.

    4. Re "fear of the fully elevated model", all you have to do is look at a couple of very nearby existing Caltrain grade separations a couple towns away.

      Howard Avenue in San Carlos isn't quite "fully elevated" (because Caltrain never does anything right, ever, even when it is easier and cheaper) but it's pretty close.

      Note that the existing roadways and pedestrian routes pretty much didn't change, the bridge is -- by Caltrain coal-hauling olde time freight railroading overbuilding standards -- pretty transparent, and this wasn't a huge road reconstruction and road reprofiling program and property and easement acquisition and impacts con job masquerading as a rail grade separation.

      Now compare to Harbor Boulevard in Belmont.

      Or, God help us all, Holly Street in San Carlos or the shitshow that is Ralston in Belmont.

      This is perfect example of the hideous "split grade separation" that Caltrain inflicts on humans because of "railroad operations constraints" (gotta design like we're a Wyoming coal mine railroad) and -- ow ow ow don't throw me in the briar patch! -- of "community concerns" about a "Berlin Wall".

      The tracks are a few feet lower. Woo hoo.

      The pedestrian environment is a mess.

      There are blocks and blocks of roadway depressions -- not just under the tracks, but on the track-parallel roads on either side. Digging up and digging down El Camino (and Old County on the other side) for many hundreds of feet isn't cheap, but hey, somebody's gotta make bank. Note that properties and walkways on the far, western side of El Camino and the eastern side of Old County were impacted and accessibility lost because of a stupid "split" road tunnel nearby.

      There was huge and unnecessary and costly utility relocation (sweet sweet sweet opportunities here ever time for "unexpected conditions" and "unfortunate budget impacts" and "unforeseen schedule slippage". They're making out like bandits on this front as we speak with the 25th and 31st Avenue roadway-project-masquerading-as-grade-separation scam at Hillsdale.

      And there's what you don't see: other opportunities for additional road and pedestrian connections across the tracks that are lost because, instead of adding a simple enough rail overbridge, you're now looking at a road tunnelling and utility digging scam with impacts to properties and roads and footpaths that aren't even adjacent to the tracks, and, well, they didn't happen and won't happen because of "cost" and "complexity".

      Caltrain is building more of these appalling "split" grade separations -- for no good reason at all, when there was no reason at all for building the right thing for the benefit of humans -- at 31st Avenue in San Mateo and coming Real Soon Now, the moronic Broadway Avenue in Burlingame massive road construction and utility relocation scam.

    5. Compare also:
      * "Split" grade separation. NOT a "Berlin Wall", we're told. No aesthetic or community impacts of any type!

      * Less-"split", higher tracks, and, one is lead to believe, a "Berlin Wall" with horrific community, noise and aesthetic impacts
      (San Bruno, which is a disaster for the usual malicious/incompetent/stupid/ignorant/rent-seeking Caltrain reasons, but not because the tracks are a couple feet higher and the roadway isn't deeply trenched.

      No human being would design anything like the underbellies of San Carlos or Belmont stations (take steps down into a dank hole order to go up to actual destination of the passenger train platform above WTF?!????!) for other human beings. It's just nasty, bad, and evil, and it's what Caltrain's "professional" staff and its perma-temp cadre of rent-seeking consultants engineer over and over and over. (Again coming to you at Hillsdale, Broadway Burlingame, and more!) America's Finest Transportation Professionals, on the job, on the take.

    6. Accordingly, there is a whole new level of complexity at Maple, where the split grade design (brought on by insufficiently aggressive grade at the Highway 84 end of things) "forces" the construction of a new siphon in place of the existing Redwood Creek culvert under the tracks.

      It's almost like the whole design process starts with "dude, wouldn't it be cool to post on LinkedIn that I designed a siphon?" and everything else flows logically from that.

  3. Dear Clem,

    Thank you for this thoughtful, thorough analysis. We have one chance to get this right and I earnestly hope that those making decisions about this project will take your input to heart. What can we do to support the uptake of this input.


    1. This sort of input usually reaches the right people, but will they take it to heart? The main point to emphasize is building for 4 tracks through downtown, because anything short of that is effectively sabotaging future rail service by permanently impairing the right of way. The more people repeating that message, the more they might listen?

    2. Speaking of repeating, who do you think we Californians should be repeating the message to in order to get your ideas implemented?

  4. Big fan of the general ideas, but if Fair Oaks station was center island, it seems like it would fit comfortably in the 80 foot right of way. Could take more width from the 28 foot platform since access would be from either end.

    1' sound wall /25' 2-tracks/28' platform/25' 2-tracks /1' wall

    1. Agreed that a center island at Fair Oaks would be even better, to enable FSSF traffic where you can turn back traffic at Redwood City (for example a Dumbarton train) without fouling the express tracks for even a second. The issue I ran into was that this would require rebuilding the Fifth Avenue rail bridge and a bunch of track, which (who knows) might have been a bridge too far. Then again... FSSF really is The Way, and island platforms rule so long as you make the tracks wow around them on a gentle curve, albeit in direct conflict with freight-centric Caltrain and AREMA design standards.

      The width would be a minimum of (1' + 8.5' + 15' + 5.3' + 14') * 2 = 88 feet, or thereabouts. You needed a bit more clearance between your sound wall and the nearest track to meet GO 26-D, and the poles and their foundations add even a bit more to that. So it ends up over 90 feet, roughly the same as the outside platform version.

  5. Shouldn’t San Carlos be built as 4-track from the start of any Redwood City overtake plan? Why should the local be sitting for 3 min at RWC waiting to be overtaken when the overtake could happen during the stop at San Carlos? If being stopped has such a high passenger penalty it would seem worth it to run the local slower than normal between San Carlos and Fair Oaks, if those stops could be the same 45 sec dwell as every other (a longer stop at RWC for transfers is ok of course).

    Why would a Dumbarton Service stop at RWC and force a transfer to Caltrain? Shouldn’t it continue on as an express to Transbay, which will of course be the destination for the majority of riders? Why force them to get off and incur a transfer penalty?

    If Dumbarton trains continue N as expresses, would we then want them to meet a SB express in RWC so E Bay riders could cross platform transfer to a train headed to Pali Alto, SJ and other Silicon Valley destinations (not as important as Transbay, but collectively significant)? To that end, is this overtake a place where trains should run SFFS, so that the Dumbarton NB meets a Express SB and vice versa? To that end, shouldn’t RWC be built (or at least designed) with a third platform between the center two tracks?

    1. You can't go north because that would cannabalize the capacity of the caltrain line. A frequent transfer is better than a terrible straight shot. People would just walk between platforms using station access walkways, no need to overbuild for a minority of riders.

    2. If you've ridden BART enough from Berkeley to SF in the evenings, you know the cross-platform transfer can be painless.

    3. It would not be a terrible straight shot, in would be a fantastic straight shot, right to Transbay. Transfers are to be encouraged, but in the context of everywhere to everywhere travel. Successful transit systems almost always provide fast direct service to the Central Business District, which has the most demand. This is especially true for commuter/regional rail like Caltrans. The RER in Paris, S-Bahn in Berlin, Crossrail in London, even LIRR/Metro North in NYC all provide service direct to the CBD (or aim to, where lines and still under construction). In many cases those systems are undoing legacy alignments that didn’t directly serve the center; we should not be repeating those mistakes and building a station that precludes or de-optimizes branching to Dumbarton.

      As to capacity, I know that the terminal design of Transbay limits capacity, but I see no reason to assume Caltrain will be limited to 8 tph for all time. If there is future capacity (or even now, I think Clem once designed a schedule with 2 tph local/4 tph express, which would leave 2 tph for Dumbarton) I see no reason Caltrain shouldn’t branch to the East Bay. Branching is a universally accepted way of matching high demand in the center to lower demand outside it; every system I mention above uses it. If there will be a branch to Dumbarton at some point (not a spur as Clem shows) it should be planned for now, which is the whole point of Clem’s excellent post. Also, as Michael notes, cross platform transfers are fantastic when transfers have to be made (I too have done so from Berkeley to SF) and so even if Dumbarton is a spur to begin, making the transfer the best possible should be considered.

    4. Unknown,
      I don't understand your distinction between a "branch" and a "spur". What's the distinction? Double-track vs. single-track? Flying junction vs. not? Something else?

    5. It is actually possible to run Dumbarton (dare we say... Altamont?) service all the way to San Francisco without stepping on 8 tphpd Caltrain.

      Start with this service pattern, with two Caltrain overtake stations at San Carlos and Fair Oaks (due to tool limitations, I substituted Atherton for Fair Oaks, but you get the idea).

      The Dumbarton trains (black trace) can slot in between a Caltrain express and Caltrain limited. No overtakes are necessary, like so, thanks to the similar speed to a Caltrain express.

    6. Kiwi,
      My definitions are:
      Branch - a service that when it meets the main line from an outlying area, joins and continues on the mainline to the center. This is what I would advocate for Dumbarton, and what Clem shows in black on his second link. Think of a branch of a tree, where the wood and veins are continuous through the trunk.
      Spur - a service that when it meets the main line from an outlying area, stops and does not continue to the center. This is what Clem seemed to be originally proposing for Dumbarton. Think of a bone spur that shoots out of a bone but doesn't naturally flow into the bone's shape.
      In the diagram at https://s3.amazonaws.com/nycsubway.org/images/maps/dallas.gif the red and blue lines branch at the north and south, while the green line is a spur with transfers at Union Station (note, in real life the green line is commuter rail service while the red/blue are light rail, and also Union Station is in the center, I'm just using the diagram for illustrative purposes).
      A branch as I define it would require some kind of junction for trains to join the mainline, while a spur could arrive on unconnected track.
      I am using these terms to represent service more than infrastructure, and make no claim that my definitions are official. In terms of track, most seem to refer to longer portions of track off of the mainline as a branch, while short sections to a single destination are called a spur.

      In your Dumbarton to Transbay service pattern, why are there two local patterns, one serving Belmont and the other not? Why not have them bother serve it (or both not) so there can be a single clockface schedule for all locals (always departing a station at 15 min intervals, instead of 14 min, then 16 min, then 14 min...)
      That said, the pattern is very good, and I believe removes most of my questions about infrastructure. It appears a northbound Dumbarton would arrive at 8:02 to meet a southbound express arriving at 8:04 (for E Bay travelers to reach S Valley) while a northbound express would arrive at 8:04 to meet a southbound Dumbarton arriving at 8:08 (so they can get back home).

    7. The two local stopping patterns are the skip-stop trains that Caltrain uses to balance speed with coverage. This pattern is Caltrain's board-adopted moderate growth scenario with a few imperfect tweaks. The original Caltrain scenario, with no passing tracks and six-minute Redwood City dwells, looks like this.

      Adding an overtaking station at San Carlos opens up enough room to shoot the Dumbarton train all the way to SF (a) without catching up to the preceding local and (b) running ahead of the following express. The slight deviations from 15-minute takt are my mistake and can probably be cleaned up without altering the basic pattern.

  6. Great post as always Clem. Absolutely agree that this should be viaduct and not embankment which would make opening Hopkins trivial.

    Of the options RWC presented, I like Option 3 Phase 1 (with some improvements) as it offers the best compromise between cost and future flexibility.

    It allows easy extension of 4 tracks to San Carlos and RWC platform length extension if necessary. I would probably close Broadway to cars to save cost, but keep open for bike/ped along with Hopkins. This would provide car crossings every 2000ft and bike/ped crossings every 1000ft which is not bad. It also defers baking in decisions south of Jefferson until some of the future conditions Clem discusses become more concrete.

    I don't see much advantage to Phase 2 of either Option 2 or 3. They could just build a full underpass at Main and Chestnut, close Maple and leave the tracks at grade. Any option for phase 2 must allow for at least 3 tracks (preferably 4) to the south to allow concurrent approach/departure from cross-platform transfers.

    I wonder if the aversion to vertical curves at bridges/overpasses is to allow shoo-fly curves there during construction? Also, the HSR spec for max platform longitudinal slope is 0.25% which all these options exceed...

  7. Going meta for a moment, has anyone been able to talk to Robert over at cahsrblog? As we head into year 13 of (attempted) CA HSR, I figured I might as well start cataloging older content unless they go down. As cahsrblog.org's domain seems to have expired, and a whois lookup implies that it'll expire in October of next year. At any rate, the comments there will probably historically significant to *someone* in 50 years.

    1. Note the URL was www.cahsrblog.com, and it is comprehensively archived (including comments) on the Wayback Machine at web.archive.org

  8. I am catching flak via private channels that in this post I am straying from the True Path that I have advocated for in the past by trying to hew too closely to Caltrain's moderate growth service pattern.

    My first sin is to have suggested a station layout for Fair Oaks with slow tracks on the outside, or SFFS, instead of the operationally superior FSSF, which gives local trains the ability to turn back at Redwood City without fouling any expresses. Other than Dumbarton turning back, why would you want such a turnback?

    This goes back to the idea of census-driven service planning, where we observed that Silicon Valley is a giant blob of job sprawl where it really doesn't make operational sense to skip any stops. So a more efficient service pattern can be constructed by lopping off the local trains south of Redwood City, turning them back north at RWC. South of RWC, there is no longer express service, just all-stops on a 15-minute takt. You basically go from Caltrain's moderate growth pattern to this "census-driven" pattern.


    1. As you can see by comparing browser tabs, as far as passengers are concerned the service quality scores are the same, but fleet size requirement ($$$ capital cost) drops 20% from 25 trains to just 20, crew costs drop 20% from 1500 to 1200 train-minutes/hour, and maintenance costs drop 23% from 1293 to 989 train-km/hour. The only question is how quickly 8-car trains every 15 minutes will fill up to capacity, at which point you might have to reconsider.

      So yes, the census-driven pattern is superior to what Caltrain is proposing, in that it does the same job with less capital and O&M.

      With locals no longer competing with the express south of RWC, Fair Oaks can be built just about anywhere you please, leaving the Dumbarton merge as the only requirement. This can be done more easily if we no longer have a need to connect to the existing four-track segment.

      To summarize: FSSF, FFS!

  9. On the topic of Dumbarton, most riders will probably be bound for mid-peninsula stops such as Palo Alto or San Mateo. So the Dumbarton trains should be through-routed with the San Mateo locals, with cross-platform transfers to Santa Clara expresses at RWC.

    Looking at all the proposals, there seems to be 3 consistent types of trains that need to be offered north of RWC:

    1. SF - RWC local, which should serve all stations (no skip-stop service) to maximize connectivity.
    2. SF - RWC express, which should stop at Hillsdale, San Mateo, Millbrae, and South SF. These trains should continue into Santa Clara County as all-stop locals.
    3. CAHSR, with one intermediate stop at Millbrae.

    Here is a schedule that allows every station to be served no less than every 15 minutes. Currently, the HSR trains are routed via SJ and Pacheco, necessitating four tracks south of RWC, but they could be easily redirected via Altamont were the opportunity to arise.

  10. This is a bit Off Topic but does anyone know what happened to the CAHSR Blog? It used to be HTTP://WWW.CAHSRBLOG.COM . Anyway, I always enjoyed being able to (in the past) click back and forth between this blog and others like it.

  11. And also apologies for anyone who has already seen but this article is interesting regarding opinion. I know that HSR from a federal level is somewhat ambitious but if it ever happened I do think that Caltrain would benefit. Of course we all assumed that with CAHSR helping Caltrain as well but here we are 12 years later....still half-invested in Caltrain on a multitude of levels (inclusive of all related rail improvements like SF TBT, Dumbarton, SJ, etc.). Unbelievable coming back from Germany and seeing how difficult in comparison it is to even get Caltrain up to speed in the Bay Area.


  12. Oh, what a surprise! The 1% grade "engineering" "requirement" and "split" grade separations are a gifts that keeps on giving, and giving, and giving and giving, and giving.

    San Mateo gets $23.8M for 25th Avenue project

    Note: this is $24 million more your US earth dollars for one single rail-over-road overpass.
    You know, because, um, utilities, and um, unexpected, and, um, community needs, and, um, rain, and um, you know, um, congestion and liveability and carbon neutral and, um, you know, stuff.

    Caltrain "engineering" is a quite literally a criminal enterprise.

    Look for more of this criminality Real Soon Now in Burlingame and -- Endless Multi-tens-of-millions of Studies And Years Later -- in Redwood City.


    1. SMCTA Grade Separation Program Update

      The letter from San Mateo’s city manager begging the SMCTA for another $23.8m highlights its best to minimize digging down and messing with utilities (and/or UPRR) whenever possible: a major point in favor of fully-elevated grade seps, minimizing or eliminating the need to lower adjacent/intersecting street, condemn/acquire properties, etc.

      Accompanying Slides on 25th Ave (3 undercrossings + new station) project overrun

      Of note:
      • UPRR negotiations impact ability to move fiber optic – Caused delay of more than 500 days
      • PG&E not originally budgeted for relocation of high pressure gas line

      Video of SMCTA meeting presentation & approval of cost overrun funding request (begins at 1:19:35).

  13. Replies
    1. I had dinner at The Habit as Sequoia Station with the kids the summer prior to last. It was a hot day, and we ate out on the little balcony.

      I was amazed over the course of 40-50 minutes or so the amount of ppl, typically younger, who traversed by from the Station to cross the street to all those large new-ish apt buildings.

      I felt that the station was finally getting used properly by patrons who walked to the station b/c housing was close.

      So I do think Sequoia Station is due for a refit, it's aged and the suburban mall pattern of stores will not meet the next few decades of growth.

  14. Nice blog. I just came across this and will now frequent this page!

    Passing Whipple recently I noted there was an online form for RWC grade separation. I just completed and submitted it, and was aghast as one of the proposals.

    There is a proposal where only Whipple is separated, and Brewster and Broadway are terminated on either side of the track.

    The study itself indicates that the congestion at Whipple will plunge to an 'F', so why include that as an option ?

    1. Every study of alternatives has a so-called "straw man" alternative that is deliberately set up to be discarded in a performative display of listening to community concerns.

    2. That makes sense...as I'm sure that question would come up as a cheaper alternative, and could be discounted immediately based upon the regressive result of the study.

  15. A teaser video of Caltrain Stadler EMU testing in Colorado. NOT my video:


    1. "high speed"? Hah.
      Does anyone know the contracted in-service maximum speed for Caltrain's trainsets? I searched, but was unable to find it.

    2. According to the Stadler data sheet (https://www.stadlerrail.com/media/pdf/kcal0220e_us.pdf), it is 177 km/h (aka 110 mph).

  16. According to https://www.railwaygazette.com/passenger/caltrain-signs-double-deck-emu-and-electrification-contracts/43002.article, the contract speed is 177 km/hr (109.93 mph; call it 110mph). So I expect they're testing somewhat faster, maybe 125 mph?

    1. It is common that stability at 10% overspeed must be proven for a certain maximum speed. So, probably in Pueblo, they will be pushed to something around 200 km/h (which is one of the speeds offered for KISSes).

    2. 121 mph will be easily reached, since this thing has a ridiculous power-to-weight ratio. 8000 kW in 7 cars, that's close to TGV / Velaro territory and probably the most powerful KISS ever made.

    3. @Max: yes, exactly my point. @Clem: I don't question reaching the speed. I'd intended to reply under the earlier "High speed? Hah" comment.

    4. OTOH, maybe with the exception of the Aeroexpress units, they are also the heaviest KISSes built so far. So, we could say that their power rating for traction and braking is "adequate".

    5. Speaking of weight, an interesting data point: the empty car shell of a short car (car E in the Caltrain consist) weighs about 16 metric tons. See photo from Twitter. That puts the amount of aluminum in a 7-car train at approximately 122 metric tons, accounting for longer D/F cars and A/B car cab structures. That doesn't seem so outrageous, does it?

    6. The Stadler data sheets (unfortunately) no longer show the net weight. That means that I have to make some guesses based on secondary numbers.

      The Caltrain KISS has a maximum axle load of 23.5 t, whereas the Swiss KISSes are at 20 t.

      The acceleration of the Caltrain KISS is stated at 1 m/s^2; the 6-car SBB KISS is at 1.1 m/s^2, and the "sports car" KISS of BLS (4 cars with 8 driven axles) gets to 1.3 m/s^2. Acceleration depends on the weight to tractive force ratio.

      That is the base for my conclusion concerning weight.

      But I agree that the blank carbody is amazingly light (I wonder how heavy the new NJTransit bi-level carbodies end up; a factor of 2 would not be a surprise to me)

    7. Interesting. The Caltrain model has the same adhesion factor as the BLS model (7 cars with 14 driven axles) so I'm not sure why the starting acceleration figures are different. In practice nowadays that's a matter of software, not just physical maxima.

    8. Yeah, software limiters do make sense. The Caltrain vehicles are heavier, but that leads to higher axle loads, leading to higher tractive force. We may have to wait and see for official weight numbers of the Caltrain models.

      One thing is nevertheless impressive: the maximum power generated when braking. We just have to hope that the grid is capable to take that energy blast…

    9. I miscounted, it's 7 cars with 16 driven axles out of 28 (temporarily until the 8th unpowered car is added, at which point it's 50%). As for the grid, amazingly, they are still now performing analysis for single-phase load cases with the electricity supplier, Pacific Gas and Electric. You would think they would have figured that out by simulation many years ago...

  17. We took Trenitalia locals from Ivrea into Turin a few times, alternately on diesel-hauled 1960s passenger trains (creaky but stylishly remodeled 20-odd years ago), and modern diesel multiple units (perhaps by Alstom). After drifting from the Alpine foothills to the Po River valley, we'd pause at one of little Chiaverano's six platforms, then reverse direction and high-tail it for Turin on the old main line, then the new main line – all the while as if there was a high speed Milan-Turin trains on our tail...which there probably were. It may be a few years premature, but it will be good for Caltrain to be able to boogie when HSR gets up and running.

  18. (OT): no CalMod monthly report for March posted yet. The monthly Caltrain Board meeting fell on April 1, and the "board packet" contained the Feb monthly CalMod report (dated Feb 28).

    1. They have always been on that cadence. Board meeting of month N contains report for month N-2. On occasion the live presentation previews progress for month N-1.

  19. The Board meeting packets, sure. But I thought the Calmod Monthly Report itself was posted about 1-2 weeks after the end of the month for which the report is dated? Am I misremembering?

    I'd expected to see the Calmod Document Library show a report about now. I'm eager to see updates on completed and projected OCS foundations. (It's like a 1930s movie serial cliff-hanger)

  20. The Calmod March 2021 report is posted! 92 foundations completed in March, down slightly from 94 in February. Caltrain continues to project foundation completion 2 months later than BBII.

    The projected foundation completion for Aug 2021 is 177, roughly twice each of the last two months. I don't see how the data supports that.

  21. The Calmod montly report for March is posted!

  22. OK, so the first thing to notice about Redwood City is that Caltrain tracks around the northern limits of the city (specifically the Howard Avenue tracks-over-road overpass a bit across the city line in San Carlos) and the Caltrain tracks around the southern limits (specifically the Fifth Avenue tracks-over-road overpass a bit north of the border with the Mighty Town of Atherton) lie at about the same elevation -- about 40 feet (feet! in 2021!).

    The second thing to note is that tracks at a uniform elevation of 40 feet -- even ones supported by elephantine Caltrain steam-train-compatible bridges with a honking soffit-to-top-of-rail depth of 1.9m -- will pass clear over Brewster Avenue, Broadway (with plenty to spare), Jefferson Avenue (at human-compatible surrounding-ground level), Maple, Main and Chestnut Streets.

    Now this simple flat elevation of feet top of rail (minus 1.9m structure depth) is few feet low to clear Whipple Avenue, but that's easily bumped up-and-over.

    But it runs right into the 1970s-era Higway 84 (Woodside Road) overcrossing of the tracks. (The elevation of the bottom of the road bridge is approximately 41 feet, the rails passing under about 22 feet. But these are not only approximate but inaccurate, since from other sources we know the minimum clearance between the exising Caltrain tracks and the existing road overpass is 23.08 feet (feet! in 2021!), not the 19 feet of clearance we'd read off the Professional Planning Documents from out special consultant friends at AECOM doin' the endless visioning thing on the public dime. (It is nice to know that the dozens of consultants at AECOM seem to be is working with the exact same super-sketch-level data I have. It's almost as if bullshit crayoning public process is just a wanking exercise before other consultants build the worst possible project anyway. Weird, huh?)

    So ... maybe ... the problem isn't how to get the tracks under the Immovable Sacred Sacrosanct 1970s Expressway Overpass ... but ... what the hell a crap overpass, surrounded by 1960s-hell suburban cloverleaf interchanges to local streets, is doing right in the way of where the train tracks obviously want to go.

    Also what the hell is this sort of hideous freeway hell doing in the middle of potentially pleasant surburban environment like Redwood City's?

    Oh, and something to think about: who's really pretty fucking good at rebuilding state highways, changing the elevation of state highways, constructing the interfaces between state highways and local roads, phasing road reconstruction while keeping lanes open, and WHO ISN'T CALTRAIN?

    (*Actually you can simply that to: who isn't worse at Caltrain at anything at all to do with either rail or road construction or operations?)

  23. "Enormous extra costs are being incurred at the southern end of the grade separation by not allowing the vertical curve at Highway 84 to begin until north of the overpass and the existing turnout, and then limiting the grade to only 1%. This is lazy paint-by-numbers engineering that blindly applies design standards without regard to their consequences, which in this case push Chestnut Street and others much deeper underground than is necessary. If there ever was a case where an exception to Caltrain design standards was warranted, this is it. Using a 2% grade and starting the vertical curve south of Highway 84, squeezing every inch of available vertical clearance under Highway 84, will save millions. In a modified Alternative 1, Main and Maple may not even need to be sunk at all."

    Using what turns out to be almost exactly the same crappy data as the Professional Visioneers at AECOM used, it's indeed trivial to see that it is possible for Caltrain tracks to pass free and clear over Maple and Main Streets and under the (100% regrettable and needs-to-be-blown-up) Highway 84 road overpass, and do so with grades that are entirely compatible with Caltrain's self-inflicted steam-hauled-freight-oriented "design standards."

    (Chestnut closes to automobile traffic, though. Can't win 'em all.)

    From the your-tax-dollars-at-work AECOM consultancy crayon, we can read off a distance of about 519m between the south edge of Main Street and the north edge of Highway 84 (all back-of-envelope; keep in mind both of these crossings are at angles to the tracks) and an existing rail elevation change of ~1.23m between those points.

    BUT ... I know for a fact the rack-to-84-soffit clearance is minimum 23.08 feet = 7.03m, while the AECOM crayon shows ~5.66m -- they've crayoned 84 either too low or too deep or the tracks too high. But you know, we're only talking ELIMINATING ALTERNATIVES from a ~$500m project, so yeah, whatever, the bridge can be as high or as low as you like, as long as the wrong thing is selected in the end. Right? Right!


    Anyway Caltrain wants 15.75 feet =4.8m of clearance between roads and railway bridges and 23 feet =7.01m between tracks and road overbridges.

    Recent Caltrain bridges constructed for the (mind-bogglingly inept and over-budget and stupid) Hillsdale and San Bruno grade separation project have bridge depths (from soffit to top of rail, including ballast) of 1.69m (18.9m spans on steel girders over San Mateo and San Bruno Avenues in San Bruno), 1.96m (25.6m steel span Angus Avenue), 1.79m (13.1m 25th Avenue concrete box girder) and 1.83m (13.7m 28th Avenue concrete box girder.)

    Let's go with 1.9m (6.25 feet) deep -- generous and in reality totally excessive.

    So we need to lose 4.8m + 1.9m - 1.23m = 5.47m of elevation in a the track run of 519m between over-Main and under-84.

    Let's plug that into the advanced Consultant-o-Matic Engineering Soft-Cost-Burden-tronic! -5.47 / 519 ~= -1.053% grade. Whoa! Perilously close to Caltrain's 1% "standard" and way below their 2% maximum, and in fact shallower than the 1.228% of the recent grade separation climb from under Highway 380 to San Bruno station platform tracks.

    Let's choose a nice round -1.25% downgrade from the elevated station (close to San Bruno's 1.228%)

    But what about vertical curves? you fret, you cry, you declaim!

    Turns out it all works out OK here in The Best of All Possible Worlds.


    1. But uncle Richard, how will they get to build the award-winning siphon structure for the Redwood Creek if everything just fits?

    2. ... Turns out it all works out OK here in The Best of All Possible Worlds.

      We're allowed -- by physics and engineering reality -- to start our descending vertical curve over or, gasp, even before Main Street. Nothing in civil engineering disallows this, and in fact Caltrain's very most recent project at 25th Avenue in San Mateo does so.

      We're allowed -- by engineering and fiscal reality -- to move the useless crossovers and turnouts of "CP Dumbarton" (junk just south of Hwy 84) a couple hundred feet in order to save hundreds of millions of road and utility relocation costs north of Hwy 84. In a normal, functional public transporation agency on a normal, functional, industrialized first-world democracy this would be the first thing to consider, rather than being a hard no-go "project constraint".

      We're allowed -- by engineering and fiscal reality -- to slightly undercut the existing Caltrain track grade south of 84 before rising again (+0.60% in my crayon) to rejoin the extant grade. Maximum undercut per my crayon (and that with unnecessarily long 50mph AREMA freight vertical curve) is less than 2.5 feet, and of course it is zero (that's the whole point!) passing under 84. And yes, it's still possible to tie in the incredibly inconvenient freight lead to the Mighty Port of Redwood City. And yes, a normal rail engineering agency could easily phase this without extensive and repeated service shutdowns. (The real difficult and stupidly constrained spot is over 22 feet above the existing awful Jefferson rail overcrossings, not two feet deeper Woodside; you can easily do the simple latter while biting the bullet on the complex former.)

      So in short:

      * Zero local road grade changes and zero blocks of multi-hundred-million utility relocation are necessary anywhere in Redwood City. If somebody else desperately wants to widen or regrade or resurface local streets, then somebody else can do that, out of somebody else's road works budget.

      * Regrettably can't go under 84 and over Chestnut Street with clearance for motor vehicles, but we can for pedestrians, which is indeed something. (If tracks go over 84 and 84 is brought back to ground level under the tracks, we can indeed do much better.)

      * Artificial "project constraints" by rent-seeking consultants designed purely to strangle the best alternatives even before "alternatives analysis" is presented for rubber-stamping result in crap outcomes at astronomical costs, outcomes so crappy it would be far better to do nothing.


  24. What year it’s going to be completed?