| High-speed networks allow the California Department of
Transportation to measure traffic performance across the entire state. |
More than 45,000 miles of highway and freeway lanes crisscross
California, and
Caltrans – the California Department of Transportation – is responsible for
managing every one of them. To measure the performance of this vast network of
roadways, the agency must determine how much delay drivers encounter, compared
to free-flow traffic conditions.
Approximately 23,000 sensors or loop detectors buried in the
freeways measure the number of cars that pass every 30 seconds, as well as the
average time that a vehicle is present over the loop. This data is transmitted
to each district’s Traffic Management Center (TMC) over a variety of networks.
Finding Common Ground
When the sensors were first installed, no centralized system
existed to collect and process the TMC data. Each district reported its data to
Caltrans headquarters using inconsistent measurement methods. There was no way
to compare data from one district to another, and therefore no unified view of
what was taking place on the freeways. Without a standard performance metric,
Caltrans could not measure and report conditions that involved two or more districts.
The lack of a centralized system also made performance monitoring labor intensive. “Historically, we’ve been doing much of the work
manually and using paper reports,” says Caltrans operations manager Bob
Ratcliff. Traffic engineers accessed data from the systems that are used for
traffic signals and changeable message signs. Caltrans also sent vehicles out
to conduct tachometer runs several times a year. Speeds, travel times and other
data were then aggregated into annual reports. “We tried to be representative
of what was occurring on the freeways [under normal circumstances], so we’d
shoot for weekdays rather than holidays,” explains Ratcliff.
In 1999, Caltrans began to centralize its processing and
analysis of loop detector data, using a new Performance Measurement System
(PeMS). PeMS was designed at the University
of California, Berkeley,
where Caltrans and the university’s Institute
of Transportation Studies maintain an ongoing
collaboration known as California Partners for Advanced Transit and Highways
(PATH).
PeMS at Work
 |
| Copyright 2001 California Department of Transportation |
To improve the measurement of freeway performance, Caltrans
began to implement PeMS in stages. The system was introduced in urban areas, such
as Los Angeles,
and slowly moved into other urban areas. At present, nine out of 12
districts collect real-time data, which are transmitted to the PeMS Oracle
database over Caltrans’ high-speed WAN. (The WAN is also used to transmit very
large engineering and geographic information files.)“Using this high-speed network, Caltrans is able to
transport massive amounts of data throughout the state to one location,” says Karl
Petty, PeMS architect and CEO, Berkeley Transportation Systems Inc. “PeMS takes
the raw lane-by-lane data and performs algorithms that allow engineers to look
at freeway performance over long periods of time. They can determine which
freeways are improving and which ones are getting worse and then decide what
kind of management strategy is needed to reduce delays on that particular
freeway section.”
Users access PeMS reports over the Internet through a Web
browser. System planners can pinpoint bottlenecks and determine whether they
can be alleviated by improving operations or by minor capital improvements. Ratcliff
recalls a manager who encountered a bottleneck at a ramp while driving. Using
PeMS, he was able to determine that the bottleneck originated in a different
location from what he had expected. “It was helpful in doing analysis and
taking that to the decision makers to show them what was taking place,” says
Ratcliff.
Recently, Caltrans added dashboard tabs, which managers are starting to use more frequently. “This is a quick
way for them to understand day by day what’s going on in their district,” says
Ratcliff. For instance, the dashboard displays yesterday’s top bottlenecks.
Managers and traffic engineers can examine the location of these bottlenecks
and conduct research into possible causes.
Another popular link is the congestion pie, which displays
in pie-chart format the causes of congestion by geographic region. “Managers,
traffic engineers and researchers use this for corridor management and
planning,” says Ratcliff. “They’ll find causality and then do everything from
looking at PeMS aerials and photo logs to driving around in order to drill down
deeper.”
 |
| Copyright 2000 California Department of Transportation |
One of PeMS’ most measurable successes has been diagnosing
faulty loop detectors. The detector health link allows managers to determine
the percentage of good loops by district and then drill down further to
determine which loops are faulty and the possible cause of the problem. Once
the information is relayed to the district, maintenance workers are dispatched
to the loops to repair them.
“Six to nine months ago, the average detector health was
around 50%,” says Ratcliff. “Since PeMS has been implemented, we’ve raised
detector health to an average of 74%.”
PeMS also receives real-time incident data from the
California Highway Patrol (CHP). In conjunction with a historical accident database,
this information is used in safety analyses.
Looking Down the Road
Since PeMS’ inception, districts have continued to fold in
their data. The most recent, District 10, began sending data to PeMS several
months ago. And while tachometer runs are becoming more sophisticated with the
use of GPS, eventually the agency hopes to phase them out completely, says
Ratcliff. The bottom line: “PeMS does a better job.” But for now there is talk
of incorporating data from the tachometer runs into PeMS – a task that would
require an algorithm adjustment, since the GPS tachometer runs capture
real-time location data every 60 seconds.
Early in 2006, Caltrans began moving PeMS in house from UC
Berkeley. The California
Center for Innovative
Transportation (CCIT) at UC Berkeley is leading an initiative to install computer
and networking hardware that will allow Caltrans to host its own copy of the
software. CCIT is also offering outreach sessions and training courses to
Caltrans personnel to promote awareness of PeMS. “We want get as many people
aware of PeMS as possible so they can use the system,” says Erik Alm, CCIT senior
development engineer. “PeMS provides much better performance data on the
highway systems, and we expect it will have benefits for everyone.”
