The MTA is Andrew Cuomo’s problem whether he wants it or not

Under CBTC and Automatic Train Operation (ATO) the space between the next train is dynamically computed based on the real distance to the next train or switch. Where as in fixed block signaling even if the last axel of train is exiting the front of the buffer block, the whole buffer block will be signaled as occupied. It is this removal of the fixed space in between trains and making it “moving” that allows more efficient use of track space and thus higher maximum TPH.

First, NYCT does not have a fixed length block system. The blocks are static (don’t move) but vary in length, depending on their location.

If two trains operate on the same speed and station dwell profile but are separated by departure time, then their separation will be at a minimum when the follower is approaching a station that the leader is in or just leaving. This is the only location and time, when the follower must know the position of the leader’s rear axle with great precision and accuracy. Block lengths approaching and within NYCT subway stations are as short as 100 feet.

The uncertainty for CBTC’s position determination must include the timeout for lost status transmissions and responses. It’s 2 seconds. If a train does not respond within 2 seconds, the CBTC system will stop all followers. This 2 second timeout, must be factored into the minimum spacing between trains because status responses can be 2- seconds late and the system should still operate. This distance traveled by a train operating at 34 mph for 2 seconds is 100 feet.

Thus, the distance measurement that controls how close followers can close in on their leaders is equivalent for both NYCT’s century old block system and CBTC. If they are equivalent, then CBTC cannot provide increased TPH because it permits reduced follower close in distances.

Looking at a letter the MTA has published from 2011 [1] they state that prior to the installation of CBTC that the realistic max TPH on the Canarsie Line was 20 TPH

Two of the MTA’s predecessors, NYCBOT (1949) and NYCTA (1954), reported operating the Canarsie Line at 24 TPH. Both agencies reported these service levels in annual reports. Here’s a link to the 1954 service levels.

N.B. if you wish to raise the key-by canard, the practice of keying-by for purposes of maintaining schedule was outlawed by the NYSPSC in 1929.

after installation of CBTC with ATO that they expect to have a higher maximum of 26 TPH with installation of additional transformers to supply more 3rd rail current. A report done in 2013 by the FTA [2] shows that during testing they were able to achieve the 26 TPH maximum

There has been no test with live trains and live passengers. The FTA Report is rather vague on what the “endurance” test was.

The move to CBTC allows elements of the Auxiliary Wayside System (AWS) to be removed.

CBTC does not eliminate all the block functions. You mentioned interlockings. CBTC does not do switching. Another function CBTC does not perform is broken rail detection. Therefore, an AWS is required. The length of the AWS’ blocks determines the AWS’ service level. It’s 3 TPH for the Canarsie Line. If a work train or a CBTC equipped train loses radio communication, the entire line reverts to the AWS 3 TPH mode. It’s the nature of CBTC. The MTA’s only contribution was deciding what the AWS service level should be.

The second new maintenance issue lies in design that the trains on the Canarsie Line use to determine their exact position and speed.

This is an important issue for a system that relies on the train to know its position.

Since a free axel (axel with no motor) was not available on the R143’s Siemens decided to use a laser based linear scan speed detector rather than a traditional tachometer.

Most AC motor controllers use a tachometer for closed loop control for motor speed. A tach output usually comes free. I’m told that the MTA’s AC propulsion system is open loop – no feedback to correct command errors. That’s the subject of another discussion about the MTA’s engineering decisions.

As a result the MTA and Siemens are replacing the OSMES system with redundant tachometers and accelerometers. For reference the Flushing Line CBTC trains use free axels which are equipped with a tachometer, speedometer and IMU to get positional information.

The stated need to remove a motor for the tachometer was to eliminate wheel slippage as an error source. It’s easy to place an external tach sensor on a motor assembly. There are shaft encoders, optical and magnetic detectors that can be placed in the gearbox. All these are removed from the elements. The problem with a tachometer for measuring distance traveled is that it relies on the accuracy of the wheel diameter. The spec is that it must be within 1/8″ of 34″. That’s 0.37%, which is not a bad mechanical tolerance for something that big.

Let’s say the train travels 1 mile, the calculated distance calculated error based on the diameter will be 0.37% of 1 mile or within 19.4 feet. That’s not good enough for the ATO to know where to stop the train. Let’s assume the train has to stop within 1 foot of its mark, then the position calculation will exceed this spec after 272 feet.

The solution is to place a “beacon” on the tracks every 272 feet that will correct the calculation back to zero error. The train’s position will be calculated by dead reckoning from the tach between beacons. I believe in practice they are placing the “beacons” every 600 feet. Given that position calibration is required so often, the free axle did not buy much accuracy.

Of course, these beacons will subject to the harsh right of way environment and will have to be regularly inspected and maintained. Eliminating the signal blocks did not end the need for expensive and frequent right of way maintenance. The number of “beacons” is probably equivalent to the number of blocks.

Accelerometers have their own error source. According to the theory of general relativity, one cannot design an instrument that can distinguish between gravity and inertial acceleration. This means that the accelerometers are measuring both the train’s acceleration and the earth’s gravity.

Inertial navigation systems try to compensate by calculating gravity from the current calculated position and subtracting it from the accelerometer reading. The problem is that the equations are unstable. The error function is an exponentially increasing sinusoid. The error isn’t a problem for short trips like ballistic missiles trajectories. Longer trips require frequent resets to keep the error within reason. The Apollo astronauts had to frequently point a telescope at some stars and enter their readings into a guidance computer. That’s why they hit the moon.

Because of the interconnectedness of the NYCT property the MTA, properly so, decided to design a set CBTC/ATO/AWS “Interface Interoperability Specifications” to assure interconnectedness between various CBTC subsystems thus allowing multiple vendors to supply different portions of the entire system after become qualified/certified. This strategy of a modular procurement is based off of the strategy the RATP in Paris has pioneered for its CBTC re-signaling projects.

There already an international standard for interconnection. It’s the ISO 7498 or simply the OSI layered model. This standard has been around since 1984. Almost the entire electronics industry supports it. It’s the reason electronic communications devices are commodities.

By writing its own, the MTA risks making a major error in its specification and/or limit interest to only those manufacturers that will humor the MTA.

The MTA and the railroad industry seems to think it’s a big market. It’s not the the electronics industry. The railroad industry’s propensity for insisting on custom equipment is the major reason railroad electronics is so expensive.

About 25 years ago, GM wanted to establish its own factory floor communications interconnection standard. The electronics industry told GM to get lost because GM wasn’t a big enough to interest them.

The MTA and the railroad industry have a much smaller presence. The result will be that the railroad industry will be served by only a few companies that will sell it equipment that hundreds of other companies sell for orders of magnitude less money. GM had enough sense to adopt the OSI model.

You seem to take issue with the choice of radio selected and link a document that recommends inductive loop over radio based communications.

No. The article compared 3 choices for radio communications: custom made; commercial off the shelf; and open standard (IEEE). The costs for same function differed by an order of magnitude between each type, with the custom radio being 100 times more expensive than the IEEE standard. The MTA chose the custom radio.

it should be noted that the result isn’t as cataclysmic as you make it seem. The QBL West Phase 1 will use the same radio’s used on the Canarsie Line.

Let’s first examine the cost implications. The custom radio costs at least $22K, as per the Tom Sullivan article. The MTA operates approximately 600 trainsets. Each trainset will require 4 radios. The cost will be $52.8M, or roughly half of what the RATP is paying to convert 5 lines to CBTC (radios included). Using IEEE radios would cost $528K.

This vendor lock-in is clearly detrimental to MTA’s bargaining capability and leaves them open to price gauging which is why they have decided to enter into long-term agreements with Siemens over the radio availability and pricing.

Obviously, one needs to look at the details, which were not presented in the referenced article. One major problem with the Siemens radio is that it does not conform to the OSI standard for interconnection. It uses its own proprietary interface that is designed for Siemens equipment This will effectively kill any interest by independent vendors that might want to design equipment for the MTA. Their sales pitch would be lower cost. The use of the Siemens radio proprietary interface will drive up the development cost for independent developers. Siemens might make its radio available to the MTA at low cost (I doubt it would be as low as an IEEE radio). Unless, it’s made available to everyone at the same low cost, its incorporation into the MTA’s CBTC will serve as a barrier to additional vendors doing business with the MTA.

QBL West Phase 1 project plans today indicate that the onboard car equipment will be Siemens based and Thales providing the zone controllers and ATS.

Thales, of course, is the old Alcatel. This is the same duopoly from the Canarsie project. I assume the reason they are getting a few more crumbs is that they threatened to drop out, if they did not. :=)

they have a third contractor, Mitsubishi, who is currently qualifying their equipment on the Culver test track to become a third vendor after Alstom dropped out.

At this point, Mitsubishi has not qualified. Even if it qualified, there’s no indication they would bid competitively on future contracts. They may come to Alstom’s conclusion, that effectively being a subcontractor to Siemens isn’t in their interest.

Belittling the agency or the technology as whole because you don’t like some of the decisions management made is unfair and unproductive.

No, my opinions are based solely on the technology’s shortcomings. The MTA’s management decisions turned what would have been a failed effort into a disaster.

If I have made any mistakes or omissions in my analysis feel free to let me know.

You have lumped CBTC, ATO and ATS (Automatic Train Supervision) together. Each is separate and can be implemented with or without the other two. The important question is which is critical to reliably achieving greater service levels?