Earlier todaythis week, we got a peak at WMATA’s plans for the future through the publishing of a presentation to be given to the board later this week. The Authority plans to take on a comprehensive modernization of their infrastructure and operating model, including systemwide CBTC deployment targeting GoA4 operation, PSDs and dynamic service allocation.

Obligatory disclaimer: While I work for a transit agency, views expressed herein are solely my own, and do not reflect that of my employer. No information included in this post is not available through publicly accessible documents.

CBTC #

Metrorail’s current train control system, though very functional, is beginning to show its age. WMATA projects significant costs to maintain the legacy system versus replacement with CBTC. In this way, the motives behind WMATA’s CBTC journey are prototypical and being pursued primarily to address an aging legacy system, with capacity improvements as an important but secondary concern.

Earlier this year, WMATA issued an RFI to vendors, in which they ask vendors to consider CBTC retrofits on the 7000 series cars, dual equipping (existing ATC and CBTC equipment), cutover strategies, PSD integration and vendor interoperability.¹ This document provides us some insight into what planners at the agency are considering, especially in terms of technical details.

We know now that WMATA intends to install and cut over the Red Line first. Given its operational isolation, this is sensible. Curiously, despite much talk of capacity improvements in the presentation, there is little talk of existing or future interlockings.

What I am most interested to see play out is the technical design of the train control system. Will there be a secondary train detection system? Will it be track circuit or axle counter based? Will there be a fallback wayside system? Where will signals be kept? To what degree will manual operation be tolerated?, possible? Will work trains be equipped?

Platform Screen Doors #

In some ways PSDs are the most interesting part of this proposed modernization program. They’re complex, expensive, present a difficult phasing and systems integration problem, and deliver benefits that I would not consider totally commensurate with their costs, financial and otherwise. But I give WMATA big points here for shooting for the stars, and if this is what it will take to sell GoA4 operation politically, so be it.

The systems integration challenges of CBTC are already great, and PSDs add significant project risk surface. I hope that WMATA can avoid some of the challenges that peer agencies looking to retrofit PSDs have faced.

My primary concern, however, is the potential for a negative impact on the ability to control dwells. WMATA already struggles to control dwells in the system (though the return of auto doors is aiding to abate excessively long dwells at stations that aren’t super busy). I worry that

1. Implementing PSDs now will lock WMATA in to their current suboptimal door configuration, and
2. PSDs will increase, instead of helping to control, dwell times.

First, on door layout: all of WMATA’s current cars have three 50 inch doors per side. This, combined with a suboptimal interior layout reflective more of a commuter rail system than a metro all contribute to excessive dwell times and poor passenger circulation. Passengers tend to hover around the doors (which are not offset), slowing ingress and egress, and decreasing effective car capacity.² WMATA looks to be interested in moving toward a more rapid transit like interior layout with the 8000 series cars, but they will not have more doors.

Installing platform screen doors without explicit provisioning for a modified door layout would, for all intents and purposes, lock WMATA in their current door count and placement ad infinitum. Increasing door count or changing placement would require expensive (if not impossible) modification of an installed PSD system, or to wait for the replacement cycle of both the entire fleet and PSDs to line up—an unlikely prospect given WMATA’s tendency to avoid whole-fleet replacement.

Dwells, again, are a problem on Metrorail. PSDs that are not aggressively tuned risk adding additional seconds and failure points to an already long dwell process. Any fraction of a second difference in door performance between car and platform is lost time, and every additional door is an additional point in which passengers can get stuck, obstruct, or otherwise delay departure. Serious conversations will need to be had within WMATA about whether this reduction in large delays caused by track intrusion is worth the many new small delay vectors introduced.

In a well tuned system, where train control, train and station all communicate flawlessly and natively, PSDs add a second or two to the door cycle. It is crucial that interfacing with PSD systems is handled natively in the CBTC system or the vendors be made to work closely. Ascertaining berthing status through any means other than directly interfacing with the signalling system will likely introduce far too much lag.

GoA4 and dynamic service allocation #

The implementation of both Grade of Automation 4 (fully driverless) operation will be more of a labor issue than a technology issue. WMATA, being the first in the U.S., will set the tone for the labor relations of the country in terms of GoA4 retrofits. There are a couple of likely outcomes here.

1. No reduction in headcount: Metro agrees to keep headcount steady. All operators are reallocated, initially to other lines, then to other duties. Savings are only realized in decreased marginal cost of service adds.
2. Reduction through attrition: Metro stops hiring new operators (or greatly slows the pace), letting attrition through retirement and reassignment bring operator count slowly to zero. Larger operational savings.
   Another interesting tidbit in the presentation is the mention of dynamic, demand based service allocation. While this is possible in a non-driverless environment, it is made exceedingly difficult and expensive, as extra crews must be kept on standby at all times at all locations where dispatchers may way to insert a train. In a GoA4 system, as long as equipment is ready, it can be deployed on short notice.