Saturday, June 30, 2012

The False Promise of Signal Timing

Oftentimes, opponents of transit will offer where they say transit money should be invested. These include things like capacity increases, new roads or highways and programs like intelligent transportation systems or other things like signal timing to reduce delays caused at stop lights. It is that last one that I want to dissect.

First, let me say what I think should be obvious in this scenario. If it were as easy as the pundits insist, it would have been done already. I have railed against the power traffic engineers have in planning our neighborhoods, communities and cities. This is a common tool they have. But there is a reason every street can't be timed.

In an urban area, the distance between intersections is very short. This is a big reason why traffic engineers love one-way streets in urban settings. It gives them a measure of control over traffic flow.

One-ways are also preferred for timing because cross streets that have traffic flow onto those streets can be controled by traffic lights. Turning from a one-way to a one-way is no different than a right on red. On two-way streets, a left turn has to cross a divergent traffic stream. This time spent waiting reduces traffic flow, which is a violation of traffic engineers thinking. That is also why there are turn lanes, but in urban areas, where real estate is scarce, that is not as easy to do.

There are other micellaneous factors as well. One of my favorites is a human behavior, and unable to be solved in a traffic engineers handbook of formulas. A common example, if a visitor is new to an area and completely unfamiliar with the surroundings, they will naturally drive slower. This conflicts with the commuter who drives in every weekday and is trying to get to the office as fast as they can. The commuter will change lanes to pass the slower driver and lane changing has proven to slow traffic. All this has a negative impact on timing because it isn't measurable.

I'll use Downtown Dallas as an example. I'll start with the Elm/Main/Commerce corridor. Elm and Commerce are opposite one-ways and Main is in the middle with a lane in each direction. We'll worry about the intersections with the north-south streets later.

Map of Downtown Dallas. I'll reference this and add crude Microsoft Paint edits. 

Starting with Elm, running west, it is easy to time them, it is one way. Taking the speed limit as the base, it is easy to measure distance and time and change the lights accordingly. It is easy to set how long the lights cycle through the green, yellow, red cycle. There are no conflicts at this point with other streets.

In a similar fashion, starting on the west side of Commerce and running east, the lights can be timed.

Both Elm and Commerce are now timed, and traffic flows freely.
No let's add a few cross streets. I'll use St. Paul and Ervay, since they are similar to Elm and Commerce. The timing here would begin at the point of intersection and work out. So at Ervay and Commerce, to Ervay and Elm, timing is measured, then using the same method described earlier, the rest of the lights are set down the street. This is repeated on St. Paul. The difficulty is low in getting each intersection timed

So far so good. All these lights are timed.
Now comes the impossible part. Main Street is two-way. If a timed direction is set at both ends of the street, running toward the other end, sooner or later the timings are going to meet. In part because the block dimensions are different, but also because we have two converging directions, Main can never be timed for both directions.

The two converging directions make timing signals virtually impossible on an urban two-way street.

Now to further hammer home the point, let's add a north-south, two-way street to the mix. The problem here is amplified from the previous example. Now we have an intersection that isn't just a one-way to a one-way, but we have a full four directions. If we had a hard time timing one street with two directions, how can another work? Which direction takes priority?

Here's the problem intersection, with no ability to time every intersection on both streets.

Lastly, let's add up just the six streets we looked at. The four one-way streets were well-timed, but the two two-ways were not. What happens though, when we add those two two-ways to the already timed intersections?

Just these six streets, without counting the others in Downtown, have no hope of being timed.

If all the other streets are taken into account, with their varying directions, different dimensions and different access points, there will never be any true timed streets. The best we can get is what we have now, some that are, others that aren't.

The last point I want to leave you with is this, if greater and faster traffic is bad for the urban environment, as I have discussed in numerous posts on this blog, why do we want to increase capacity and speeds on urban streets. Shouldn't the idea be to slow down traffic, thereby making pedestrians feel safer and therefore more likely to have a vibrant urban core?

1 comment:

Ken Duble said...

Thanks for explaining this. It's something I've always understood intuitively, but I've never heard anyone express it in such a concise manner.