Is there any rule -- NORAC or CFR -- which actually bans the use of railroad signals being shared with vehicular traffic signal cycles? It's rather hard to dig up. I figure if the host railroad is willing to do it, then why can't it be done?
And if its a matter of track designation, then plop a station sign 100 ft west of Mass Ave and 100 ft east of Broadway. Make that other-than-main-track between the station signs and add traffic signals. Throw in some special instructions in the employee timetable if you have to. I'm not convinced there's no easy way around this. Seems to big of a deal over a simple issue.
FRA's website was designed with search functionally straight out of a civil servant IT guy employed during Clinton's second term. For the most part Wikipedia and other Google-able sources list the essentials if you don't feel like pulling your hair out searching through 50 pages of poorly-sorted public records.
FRA regs, the NORAC et al. train crew rulebooks, and the MUTCD road design manual (which is much easier to search) all converge over crossings.
The rules are:
--
[FRA rule] Active crossing protection must be triggered a minimum of 30 seconds before a hypothetical train hits the crossing while running at max authorized speed (MAS) for the segment of track surrounding the crossing. It's not a variable trigger at different speeds because it's based on MAS, therefore you get the same full-speed timings for a nonstop train even if 100% of the trains are slowing into a station stop at the crossings.
For example, with the Fitchburg Line upgrades the MAS end-to-end was changed (with only a couple small locally-restricted exceptions) from 59 MPH to 79 MPH, so every single grade crossing had to be re-timed the whole distance from Somerville to the last outbound crossing in Shirley. Ditto the Knowledge Corridor for the Vermonter when that went to 79 MPH. Ditto the Cape Flyer to a MAS of 59 MPH on all the upgraded mainland track, and on parts on the Cape where speeds didn't increase nearly that big but did outstrip the old gate timings (in which case they got re-done for future 59 MPH and just run gates-down a little long today). It's not particularly difficult to re-time a crossing, but it is totally a hard-wired thing with fixed trackside position. This isn't one of those things like PTC where you can have variable spacing auto-adjusted by software. Nothing has to be more fail-safe than crossing protection, so their attitude (which has some merit) is: don't ask us for special waivers and Jetsons-shit software, because paying the going rate to grade separate is the fail-safest of all.
'Nimble' accel/decel on a particular vehicle makes no difference, just like making a station stop makes no difference. Because the timing is fully equipment-neutral and neutral to
actual vs.
max speed, the only thing that matters for gate timing is distance to the crossing vs. MAS for an any-vehicle. So, a
hi-rail pickup truck lollygagging past the Miner Ln. crossing on the NEC in Waterford, CT while doing a routine track inspection between Amtrak slots would trigger the gates as far ahead of the crossing as an Acela or Regional or Shore Line East train doing the 90 MPH passenger MAS in that area, or a P&W freight doing the 80 MPH freight MAS in that area. No exceptions; it's all distance-based, zero vehicle-based.
--
[FRA rule] Gates must be triggered by a track circuit when available. So the way that would work is that hypothetical 30-seconds distance at MAS out the train's (or hi-rail truck's) wheels take an electric pulse through one of the running rails and completes the 'circuit' through the steel wheel to the other running rail. Detector triggers the gates upon completion of the circuit. Gates stay down until the train clears the
other side of the crossing, then takes the pulse from the track circuit situated on that other side of the crossing to complete the circuit that switches the crossing protection off. The on/off circuits can detect which direction the train is moving by which running rail is making the circuit, so that's how the gates can go "off" the second the train clears in any given direction and not have to wait a matching "30 seconds out" in any given trailing direction.
Variances: when track circuits aren't available (like on a crap freight branchline) they still use crude old mechanical switches to turn the crossing on or off. The line also doesn't have to be signaled at all to have a track circuit. Insulated rail joints mean the circuit can be made localized to whatever lone stick of rail is 30 seconds out, and wired into the crossing power supply box hooked up to the nearest roadway telephone pole. With limited battery backup if there's a power outage on the local grid. All fully isolated, RR doesn't have central control or monitoring. Sometimes the RR doesn't install them at all and it's just the DOT requesting phoning in permission from the line owner to install it themselves. As is common practice, the crossing protection can trigger a nearby traffic light for traffic management while the train is passing and queue management after it's passed. It's unidirectional, though...train gets 100% of the right of way and traffic lights are reactive to the train but not vice versa.
Usually when you have a crossing malfunction where the gates get stuck, a track circuit malfunction is to blame. Like, for example, the "off" circuit didn't get detected properly so it just stays in the "on" position forever until somebody does a manual reset. Or there's a timeout phase if they're in the "on" position much too long for there to plausibly still be a train nearby. Or the train has completely stopped short of the crossing. e.g. It's a work train that stopped to do some adjacent track work...or there's a freight turning out onto an adjacent siding
after triggering the protection, but it'll never actually reach the crossing. And there are some limited things they can do around station stops to keep the gates from staying down during a full station dwell. There's a couple such installations on the T, and plenty more that badly need it. But leeway is limited, and likewise distance + MAS vs. not moving at all for X duration related...not vehicle-related and not optimization-related for the very existence of station stops with some trains.
^^The setup of all of this around distance to the crossing vs. MAS is why it makes zero difference what vehicle you're running or how nimble it accelerates/decelerates. It's fixed, hard-wired track hardware that doesn't care what the vehicle is, and by-design doesn't make the slightest attempt to care. The only thing it makes the slightest attempt to care about is which direction the train is moving for the on/off triggers. Mainly because it's easy and fail-safe to detect direction (no additional complexity, impossible to fake direction from the track circuit's perspective).
I repeated it in every other thread...it don't mean shit what an xMU is technologically capable of doing vs. a big dumb freight train or push-pull when crossing protection is (intentionally) made blind to vehicle type. And hypothetical future awesome-waivers for running awesome-DMU's don't change anything with the trackside hardware. You either spend 2x the cost for 2x the crossing hardware at every crossing like the RiverLINE and other time-separated lines, or give up most of the flexibility. Because this is a fixed hardware thing, not a bureaucracy thing or a software thing. You certainly can't switch on-the-fly between 2 sets of completely separate gate-timing hardware installs at anything resembling fast. The self-diagnostic testing for each switchover would eat up several minutes just checking itself for fail-safety. Like doing a 'cold reboot' of the entire Grand Junction every time a conventional train and DMU are in adjacent slots. That's not quick or nimble. And if something faults on self-check it's a schedule delay while Central Ops checks it out. Several times a day if one crossing tends to fault a lot more than the others (winter weather?).
--
[NORAC, etc. rules] Horn + bell sequence must be sounded between 15 and 20 seconds, and no less than 15 seconds, before passing through the crossing. And continue both until the train has entirely occupied the crossing (i.e. additional horn + bell sequences if the train's moving slower than track speed). If it's a quiet zone, then there's no horn but still mandatory bells. And there's also the alternating flashing ditch lights rule, although that starts much further out. This all is how the crew rulebook(s) account for the varying speeds their varying rail vehicles would travel through the crossing; repeat the warning sound sequences for however many times it takes to get there if they're running slow.
The rulebooks also have lots of general-purpose stuff about crossing safety. Including the rules about stop-and-protect and flagging unprotected public crossings that have no warning devices except for a crossbucks sign. And other stuff in their route qualifications about memorizing what crossings are public (always stop-and-protect) vs. private crossings (never stop-and-protect) vs. high-traffic private crossings (waiverable into stop-and-protect rules). And on protected crossings they've got all sorts of fallback rules for dealing with malfunctioning crossing protection.
However, it's not the train crew's responsibility to watch the gates go down. They can't see the flashers and gates from their line of sight, so that's the autos' responsibility to look both ways, not blare their music so loud they can't hear the horn/bells, and obey the "No Train Horn" warning sign at quiet crossings. 30 seconds out at MAS usually isn't enough to completely stop a long train to begin with after throwing it into emergency. And it is possible for obstruction detection sensors to make the signal system
automatically send the train into emergency; this is how the NEC's crossings are set up. But the rulebook is equally clear on what's totally outside the train crew's control and the car's/biker's/pedestrian's fault.
--
[MUTCD rules] Road warning signage, safe road sightline regs, traffic densities where active protection strongly recommended (usually non-binding for upgrades, but new construction has to follow 100% of the rules). And the stuff that's the driver's fault for not obeying.
MUTCD at least has a slick website for pulling all this stuff up quickly and easily. And they have lots of pictoral illustrations of the rules that laypeople can understand, unlike the gobbledygook walls of text the FRA has. If you can locate it at all on their horrible website.
Light rail is completely different. And so is time separation when the line is exempt from FRA rules and can impose alternate gate timings when it's on exempt hours (granted: no interconnections to 'on-duty' FRA tracks allowed, so that's the killer for any RiverLINE-type applications in New England). All you have to worry about with light rail are the NTSB's much looser "don't be fucking idiots risking safety and you can do what you want" regs and whatever the MUTCD specs for road-related sharing. BRT and busways tilting much more exclusively to the MUTCD, of course.