Tuesday, September 26, 2017

Why Baltimore's traffic signals can be so much trouble

Transportation has moved to the front line when evaluating metro areas for quality of life, equity and a competitive environment.  Discontent with area transportation in general quickly leads to the criticism how transportation is managed, run and planned. Sometimes, it is worth to look at details to understand what the systemic problems are. Today we are looking at Baltimore's traffic signals.
An early Baltimore manual signal
The traffic tower at Charles street and North avenue, with Patrolman Adolph Von der Linde at the dials, was put in service on the evening of March 9, 1922, with appropriate ceremonies. This was Baltimore’s first traffic tower, copied from New York and Detroit. On each of four sides it had three large lights, red, yellow and green. It was, in short, the last word in safety. And it was not struck by an automobile until seven days later, when an unknown motorist removed part of its concrete base. (Sun archives)
Signals are in the news again for the cameras that ensure people obey them and whether those cameras are accurate. But what about the signals themselves? Are they doing their job? Many think that signal management in Baltimore leaves a lot to be desired. What is the job of a traffic light?

I recall when in the early sixties German cities started to implement "the green wave", meaning that on major arteries all traffic "lights" would be synced in such a manner that a platoon of cars could move over a long stretch and always hit a green light, intersection after intersection. Go too fast and you would be caught on red, go too slow and you would lose your platoon. There was something magic to this, even though today we don't share the goal of removing all obstacles for free flowing cars and we don't like the one way streets any more that were a direct consequence of the signal timing for free-flow idea. In Baltimore it was Henry Barnes who came to the city in 1953 and implemented these very same strategies possibly ahead of Germany which was then still a backwater. According to lore, Barnes installed in 1957 the world's largest traffic control computer right here in Baltimore. Of course, at that time a single main frame computer filled a classroom but had less operating power than Apple's very first I-phone.
First Baltimore signal tower at Pratt and Light: Looks chaotic

But even today, in the age of multi-modal "complete street" considerations and active transportation with an emphasis on walking, biking and transit, signal timing remains a valid topic. There is little benefit in stopping traffic randomly at every light simply because the signals aren't timed properly. The resulting back-ups create pollution, frustration and tempt impatient folks to blow through the randomly switching lights. In short, mentioning bad signal timing isn't propaganda for a more car friendly city, instead it raises the question why a relatively simple thing that worked in 1960 doesn't work in current day Baltimore. Obviously, Barnes didn't need fancy satellite signals or global positioning to create a "green wave". All that is needed timers and a program that establishes the optimal sequence derived from a test vehicle traveling in the desired speed. The most advanced element at the time was a special signal that indicated at which speed the "green wave" would be met, giving folks an incentive  not to speed.

There is some evidence that things don't works so well anymore because the system was gummed up with new more elaborate elements such as vehicle detectors, a central computing unit much more complicated than 50's technology permitted. Detectors can sense actual vehicle and ideally adjust the time of signal phases to actual volumes. When signals sit in a row such adjustment happens within a fairly narrow window defined by synchronization with neighboring signals. vehicle actuation allows signal phases for designated left turns and the like to be skipped or extended as needed. The bad news: Those pavement embedded detectors are vulnerable and frequently fail making all the extra phases go through a fixed cycle no matter whether there is demand or not. Worst, there are no metrics when and where to install detectors or a reliable data base where they are. Experts think that in an urban setting with a dense sequence of signals those adjustments via detector are not only a waste of money but often even a bad idea. Outdated old technology and new elements seem to have created a unholy hard to manage union. As a result the following things seem to occur in Baltimore on a regular basis:
Ancient signals still dangling over Baltimore's streets
(photo Philipsen)
  • Signal timer creep which happens when individual clocks manage the signal timing. If one of them is a few seconds fast every day, even the most perfect sync will fall apart after some time.
  • Vehicle detectors fail on a regular base and make signals run through a fixed cycle which is worse than without the the activators because the signal will run through the added phases it would likely not have without the promised efficiency of detectors.
  • There are no standards for when and where to install detectors nor seem to be even complete records as to where they are installed
  • Most of Baltimore's signal equipment is old, often very old. One can find signals in the City which date back to the 1950's. Controllers and computers become obsolete faster than the other hardware. The more the city moved away from the simple analogue systems of the past, the less reliable the system became. 
  • Many signals are now managed from a central control room the central computer "loses signals" that it previously recognized, mostly when they are peripheral to the zone the controller manages. 
  • New equipment is expensive, complicated and requires proper operator education and operation procedures which seem to be lacking. 
  • New demands, for example for TSP, transit signal priority, add new complexity requiring that different agencies use compatible software and hardware that can, indeed, communicate to the extent that a bus can influence the signal phasing (a bus will always fall out of sync with the platoon of cars since it will have to stop for passengers, ideally right behind signals and not, as frequent in Baltimore, right before them).
Before signal experts or BC-DOT officials jump all over me for for maybe being slightly off on some of the details in the above list, let me explain that the list wasn't obtained from a secret dossier or from sitting in an outdoor cafe overhearing DOT engineers at the next table, but is the result of observations, some research, consulting on traffic planning and conversations with BC-DOT staff about particularly badly behaving signals. Those conversations go way back when Frank Murphy was in charge of these matters, a stalwart traffic planner that has been around at DOT almost as long as the signals themselves. Timing signals in the Howard Street corridor so that trains would go a bit faster took 25 years before it finally worked well this year.

Also, BC-DOT has been on the matter of signals for years (modern controllers have been in the works since 2004 and in 2008 when  presumably a milestone had been reached. But 2008 was one year after introduction of the first iphone, a device long considered  obsolete now. In spite of that, progress has been made, even though increased complexity was occasionally detrimental to optimal operation. As a result progress is hard to detect for people who see only the results and not the effort.

BC-DOT describes the responsibilities of its signal engineering section on its website this way:
The Signal Engineering Section is responsible for the design of new and existing traffic signals. Specific projects include:

-Development of a state-of-the art Traffic Management Control Center (TMC)

-Replacement of the City's 1300 Traffic Signal Controllers

-Replacement of the 40,000 plus traffic signal bulbs with LED's (Light Emitting Diode)

-Signal timing contract to retime signals in the Central Business District and along selected gateways

-Installation of traffic monitoring cameras

-Installation of fiber optic cable to link signals and cameras to the TMC
That is a lot of stuff, especially since the City lost 35% of its residents but hardly any signals. And it doesnt even include signals for pedestrians and bicycles and the specific concerns that come with those "active modes" such as not enough green time for the elderly to cross, no signals at all for bicyclists going counter flow ("use the ped signal"), non functioning ped push buttons and the like.

All issues with signals may not even be all that important all by themsleves given all the other worries in this city, except they can teach valuable lessons which are applicable to other departments and even private business. If those relatively simple things don't work right, trust in the system becomes undermined. Some of the lessons in random order are:
  • Don't make things more complicated when simple works as well. (like four way stops)
  • Don't opt for complex technology if you can't afford it in the first place or don't have the capacity to operate or maintain it
  • Don't fall for vendors and consultants that always want to sell stuff but don't care how it can be managed afterwards 
  • Establish strict metrics for when and where what technology is appropriate, ensure those are met and that records are kept where which technology has been used
That the signals need cameras to monitor motorists and issue citations if they don't stop highlights the crux of technology: The efficiency of the automated cop needs to be augmented by an automated enforcer. Efficiency through automation is often not what it is touted to be.

One last word on time wasted waiting for a light to turn green. Most signal cycles don't exceed 2 minutes and 30 seconds, most less. Typically 3 minutes is considered the maximum except for special conditions where programs are overridden to "flush out" event traffic, for example after a game. Two to three minutes can seem very long, especially if one is in a hurry. But then, if people would stop reading their phones when lights turn green wasting precious seconds, things would work much better. Flaws in traffic flow usually have cumulative effects. Automated vehicles would take the human factor out and possibly make even cutting edge signal technology obsolete.

Klaus Philipsen, FAIA

A photo caption was updated

Baltimore County FAQ on traffic signals
The History of Traffic Lights in Baltimore

Related on this blog: Why downtown traffic is such a mess

Baltimore traffic management - frozen in time


  1. My opinion of Baltimore's "Democrat lights" is they are synchronized to increase motorists' stops to gain revenue from increase gas consumption. How many lights will turn green, and a block away the next light turns red? Democrat lights...

  2. Your supposed "North Avenue" photo either shows Pratt and Light at Inner Harbor instead, or some VERY high water in the background......