Other ways to use high resolution data

1 post in this topic

Posted (edited)

Clark County has installed 2070-1C CPU's with Trafficware firmware versions including high resolution data gathering at each of our intersections.  Each of these intersections communicates with our central system via high speed Ethernet.  We are in the process of establishing performance measures using the tools available in our ATMS system along with other information such as travel time from Bluetooth readers, permanent count stations etc.

One comment that we receive is to the effect of "I waited forever for a green".  This is especially an issue if the citizen is driving late at night.  We have retrofitted all of our signals to include stopbar detection on all approaches so that during low volume times, we can program the signals for either short min greens on the main street, or in many cases, placing the main street into soft recall.  So when we receive a comment that someone waited a long time during the middle of the night, we want to know if this was real.

The high resolution data can be mined for some information about delays on the side street.  The example below is for a signal that is at the exit from a County facility that includes a fueling station that the Sheriff's office uses during all hours.  One deputy called and complained that "the signal always takes forever to get out of the facility at 10:10 PM".  Now, from an engineer's, "forever" means a very long time - as in "forever".  I don't know how long the deputy thought he should wait, but it was worth looking into, especially since we have the data available, as long as we rethink some of the metrics that are in the Purdue coordination diagram.

Here is a screen shot from our Purdue coordination diagram for a short period of time late at night from one of our signals.  The red circles show the northbound left, and the northbound thru / right lane stopbar detection actuation in relation to the beginning of the green.  This seems to indicate that there is a pretty snappy response from the signal.



Here is the opposite approach (southbound stopbar detection) approaching the intersection at the same time and date.


The red circles are where there was a detection and service.  The blue circles are where there was no detection, but there was a service.  The blue circles in this correspond to the northbound detection calling the phase.


Note that one cycle did have a long wait, something on the order of 35 seconds or so at approximately 10:03 PM




Here is the main street approach where both Eastbound and Westbound detection is shown, to attempt to see if there is a gap in the traffic causing the longer delay for the 10 PM car delay.



Is there a problem here?  Maybe not.  It looks like there are cars on the main street (the black dots are both eastbound and westbound, so we don’t know how the gap out procedure is working in the controller when the one car had a long delay).


We have looked at several instances where people report that they wait “forever” for “no cars at all on the main street”.  The high resolution data helps us understand what is happening.  Sometimes the citizen is right, sometimes, the data doesn’t support what the person is saying.  I choose not to argue with the citizen, but this helps us understand if there is a problem that we could help fix, or not.


Here is another example where there was a problem we found by poking at the data.  The signal was in MAX recall on the main street, not min or soft recall, so it was holding with no cars.  It may be that we had a maintenance activity at this location, and the signal was placed in MAX recall for the maintenance, but not placed back into MIN recall.  Surprisingly, we heard about this from a citizen who went through late at night, and he contacted us on February 22.  We did a little digging, and found out he was correct.  We changed the main street to min recall yesterday afternoon, and he called the next day and was extremely happy that he could get through quickly now.




Like I said, this isn’t the standard way of using the Purdue methodology, and I wouldn’t want to stake my reputation on it, but it is a nice way to see if maybe things that we hear from citizens are normal, or not.









Edited by Rob Klug
1 person likes this

Share this post

Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in

Sign In Now