Appendix A: Performance Characteristics Analysis
A.1 ICM Factors Contributing to Achievable Reduction of Traffic Congestion and Travel Time
Users of the I-880 corridor experience significant congestion during all time periods. The table below summarizes congestion trends for three years for the freeway portion of the corridor.
| Year | AM Peak | Mid Day | Evening and Early AM | PM PeaK | Total Daily |
|---|---|---|---|---|---|
| Northbound Direction | |||||
| 2003 | 1,499 | 1,237 | 552 | 2,547 | 5,835 |
| 2004 | 1,124 | 1,067 | 360 | 2,317 | 4,867 |
| 2005 | 1,331 | 1,434 | 285 | 2,351 | 5,402 |
| Southbound Direction | |||||
| 2003 | 1,924 | 1,397 | 276 | 2,249 | 5,846 |
| 2004 | 1,728 | 1,427 | 291 | 2,375 | 5,821 |
| 2005 | 1,678 | 1,848 | 232 | 2,444 | 6,202 |
| Total Corridor | |||||
| 2003 | 3,423 | 2,634 | 828 | 4,796 | 11,682 |
| 2004 | 2,852 | 2,494 | 651 | 4,691 | 10,688 |
| 2005 | 3,009 | 3,282 | 517 | 4,795 | 11,604 |
Of course, travel time and congestion are strongly correlated. When congestion is high, so is travel time. When congestion diminishes, travel times are reduced. Whereas transportation practitioners sometimes focus on total congestion as measured by aggregate delays, individuals (i.e., customers) focus on travel times. This is an important distinction, since a 10 percent reduction in delay often translates into one or two minutes of travel time reductions. Although such reductions do not always seem significant to individuals, they do reflect impressive performance improvements overall. Over time, without such improvements, individuals would eventually experience significant travel time increases.
The exhibit on the next page depicts the travel time distribution on the I-880 corridor in the southbound direction for 2005. Notice the two peaks during the AM and PM peak periods respectively. Also note the variability of travel time over the year. Although the mean travel time is around 46 minutes during the AM peak period, many travelers experience much higher travel times. So when travel times are analyzed, it is important to consider both the mean travel times and the overall distribution of travel times.
Figure A.1 larger View - use back button to return.
(Source: PeMS, https://pems.eecs.berkeley.edu/)
The proposed ICM strategies are estimated to reduce congestion as follows:
- Reduction of overall delay by 10 percent
- Computed as total hours of delay, broken down by facility, mode, and bottleneck area
- Peak period contraction by 10 to 15 percent
- Computed by bottleneck area as the length of the period where speeds are below 35 miles per hour
- Improvement of reliability by 10 percent
- Computed by the travel time index (the 95th percentile of travel time distribution)
These are admittedly lofty goals, especially on a congested urban freeway. However, our experience suggests that these goals are attainable. These goals only translate into 2-3 minutes in mean travel time reductions, but much more for some travelers. Implementation must take into consideration several critical factors:
Performance measurement – implementation of the ICMS is expected to be iterative. To fine-tune the different systems and business processes, continuous measurement is critical. Given the detection in place on the freeway and critical arterials, practitioners will be able to evaluate their actions in almost real-time.
Focus on known bottlenecks – congestion reduction has to focus on the causes of the recurrent and non-recurrent delays. Recurrent delays are caused by known bottlenecks already identified by the region (refer to ICM Concept of Operations for a more detailed discussion of known bottlenecks). Many of these bottlenecks relate to high traffic flow merges at several locations. Strategies such as the integration of arterial signal and ramp metering systems should focus on these bottleneck locations in order to achieve significant improvements in overall congestion.
Cross-training – integration of existing systems represents the technical aspect of the ICMS. However, to properly utilize integrated systems, practitioners in charge of one system must be trained to better understand the other systems to better understand the ramifications of their actions. For instance, ramp metering staff should be trained on the systems that control the signal timing of the arterials, and vice versa. Such training will be included in the implementation plans for all ICM strategies.
Demand Management Monitoring – The proposed ICM strategies address traffic management (e.g., arterial signal integration with ramp metering), traveler information (e.g., travel time estimates for trucks at the ports), incident management (e.g., faster detection of incidents and causes thereof), and increased transit ridership (e.g., signal prioritization). All these strategies, to some extent, deal with demand management. Understanding changes in demand characteristics will therefore be critical to properly implementing the ICM strategies. Demand management metrics to monitor will include: flow rates, truck volumes under different conditions, average time to clear incidents, transit ridership, ramp waiting times, and average bus speeds. The ICMS implementation team will develop a detailed monitoring plan for these and other pertinent metrics to evaluate the effectiveness of these strategies.
Risks to Implementation
The primary risks to implementing the ICM strategies can be summarized as follows:
- Technical – integrating multiple systems requires careful analysis and detailed functional specifications
- Institutional – Systems do not improve performance on their own. Business processes have to be revised to incorporate these new capabilities. Agencies that focus only on "their” system and refuse to consider overall mobility improvements will not achieve the desired results.
- Funding – too many transportation technology projects under-estimate the total funding requirements and end up reducing the true functionality needed to improve system performance.
The remainder of this section focuses on specific strategies that aim to improve incident management, mode split, and safety. Note that these strategies all contribute to congestion and travel time reductions as well.
A.2 ICM Factors Contributing to Balance of Demand Freeway and Arterial Highways
The I-880 arterial corridor network, approximately 45 mile long, has 174 traffic signals and 15 metered ramps among some 60 on and off ramps. A preliminary analysis has been performed by the I-880 ICM team to evaluate the potential benefits of coordination between freeway ramp metering and arterial traffic management. The network was coded in NetZone , and evaluated for three control scenarios in a 3.5-hour peak period:
- Base case: the original control schemes imported from the Paramics traffic control files (ramp control is pre-timed)
- Base case without ramp metering: we kept the control schemes for intersections established in the base case but disabled ramp metering.
- Base case with ALINEA ramp metering: we used the local, traffic responsive ramp metering algorithm ALINEA to control the 15 metered on-ramps. The control gain in ALINEA is set to 70 veh/hr.
In all the three cases, we also introduced various levels of intelligent route guidance:
- no route guidance (0%), where all travelers would choose their free-flow shortest path(s).
- 100% route guidance, where all travelers would periodically re-route to their current shortest routes based on updated travel times
- 50% route guidance, where half of the travelers would follow the fixed shortest paths obtained from free-flow travel times and half of the travelers would acquire real time traffic information and periodically reroute based on the updated travel times in the network.
The primary results of network performance are shown in Table a.2. It is clear from these results that real-time traffic information, when distributed discretionally, is most effective in improving corridor performance (up to 34% travel time savings), while for the same level of information provision, ramp metering can generally improve corridor performance, but not to the extent obtained from route guidance. Signal coordination, which could further improve network performance, is not tested here due to limitations in the current version of NetZone.
| Scenarios | Total travel time (hrs) | Total travel delay (hrs | Avg. travel time (min.) | Avg. travel delay (min.) | Avg. travel speed |
|---|---|---|---|---|---|
| 0% | 1 148083.48 | 103809.74 | 39.13 | 27.43 | 19.84 |
| 2 140564.74 | 96291.01 | 37.14 | 25.44 | 20.90 | |
| 3 141367.15 | 97093.42 | 37.35 | 25.65 | 20.78 | |
| 50% | 1 98156.98 | 53365.14 | 25.94 | 14.10 | 30.07 |
| 2 99313.12 | 54455.49 | 26.24 | 14.39 | 29.70 | |
| 3 98462.97 | 53650.06 | 26302 | 14.18 | 29.96 | |
| 100% | 1 116998.11 | 70215.46 | 30.91 | 18.55 | 25.76 |
| 2 120477.39 | 73511.47 | 31.83 | 19.42 | 25.05 | |
| 3 122515.10 | 75788.03 | 32.37 | 20.03 | 24.60 |
A.3 ICM Factors Contributing to Achievable Improvements in Incident Detection and Efficient Incident Management
Based on the TASAS accident database and California Highway Patrol (CHP) logs the average number of daily I-880 incidents is over 100, with many of them occurring in the afternoon, corresponding to high traffic volumes and high levels of congestion. The chart in the following figure shows number of incidents by time of day based on a one-year sample of accident data on I-880.
(Source: PeMS TASAS)
Figure A.3 Statistics of Incidents for I-880
Based on CHP logs, incidents that most likely impact traffic represents 70 percent of all I-880 incidents reported. The chart in the following figure shows number of traffic-impacting incidents by time of day on I-880.
(Source: PeMS TASAS)
Figure A.4 Traffic Impacting Incidents on I-880
Between 5 and 15 collisions occur daily on I-880. These collisions are major sources of non-recurrent congestion on I-880 depending on the time of day these collisions occur, on the number of vehicles involved, and on the number of lanes blocked. The following figure shows number of collisions on I-880 from 1999 to 2004 based on TASAS data.
Figure A.5 larger View - use back button to return.
Figure A.5 Number of Collisions on I-880
The current incident management operation on I-880 includes: a) the Freeway Service Patrol (FSP) provides incident response 6-10 AM and 3-7 PM, and b) BAIRS (Bay Area Incident Response System) provides incident response outside of FSP hours.
The objectives of ICMS include improving incident detection, verification, response, and clearance. Possible ICM improvements in the area of incident management include:
- Improve response time for supervisors to come to incident scene and verify conditions
- Provide special heavy-duty incident response trucks (either roving or stationary) to assist with removal of trucks or in case on multi-vehicle accidents
- Improve accuracy of incident logging at CHP CAD system
- Improve integration between freeway and arterial operations
- Improve integration between freeway and transit operations
- Improve incident information including location, number of lanes blocked, etc.
Range of expected improvements in incident detection and efficient incident management resulting from ICMS deployment
The proposed ICM strategies are estimated to reduce incident–related congestion as follows:
- Reduction of overall incident delay by 10 percent
- Computed as total hours of incident delay, broken down by facility, mode, and bottleneck area
- Improvement of travel time reliability by 10 percent
- Computed by the "buffer index"
The following table summarizes observed improvements in incident detection and efficient incident management, throughout the country. Sources of this information are also provided in the table.
| Expected improvements in incident detection and efficient incident management | Source |
|---|---|
San Antonio - System decreased incident response times by:
|
from ITS Benefits: Continuing Successes and Operational Test Results - Mitretek |
Philadelphia (TIMS):
|
from Draft ITS Benefits: 1999 Update, March 1999, Mitretek Systems for FHWA ITS Joint Program Office |
Houston (TranStar):
|
from Draft ITS Benefits: 1999 Update, March 1999, Mitretek Systems for FHWA ITS Joint Program Office |
Northern Virginia - Estimates of reduction of incident duration for all incidents is:
|
from Incident Management and Intelligent Transportation Systems Technology: Estimating Benefits for Northern Virginia, March 1998 - G. Maas, M. Maggio, H. Shafie, and R. Stough, George Mason University |
National (urban) - A reduction in incident notification time from 5.2 minutes to:
|
from ITS Benefits: Continuing Successes and Operational Test Results - Mitretek |
Risks to Implementation
Risks and factors from a technical implementation standpoint include:
- Integration of Fire Department CAD system to the 511 and East Bay SMART Corridors program
- Installation of Mobile Data Terminals (MDT) on all fire departments apparatus
- Improve accuracy of incident logging at CHP CAD system
Risks and factors from an institutional implementation standpoint include:
- Logistics of coordinated response among agencies for incident management will require detailed interagency agreements
- Better coordination between supervisors responsible for incident verification
A.4 ICM Factors Contributing to Achievable Improvement in Safety
In addition to improving delay and the reliability of travel time, proposed ICM strategies are expected to also improve safety on the I-880 corridor. The proposed ICM strategies are estimated to improve safety as follows:
- Reduction of overall number of crashes by 10 percent
The following table summarizes observed improvements in safety, throughout the country. Sources of this information are also provided in the table.
| Expected safety improvements | Source |
|---|---|
| National - Accident savings of 42% for deployment of basic metropolitan ITS infrastructure (297 metro areas) | from ITS National Investment and Market Analysis, 1997 - Apogee and Wilbur Smith |
Accident statistics for fatalities:
|
from Intelligent Vehicle Highway Systems Operational Benefits – Mobility 2000 - March 1990 |
| Ramp metering reduces freeway accidents by 20 to 40% | from Intelligent Vehicle Highway Systems Operational Benefits – Mobility 2000 - March 1990 |
San Antonio - System decreased:
|
from ITS Benefits: Continuing Successes and Operational Test Results – Mitretek |
National
|
from ITS National Investment and Market Analysis, 1997 - Apogee and Wilbur Smith |
| National - Freeway management combined with incident management programs can reduce accidents by 15% to 50% in congested areas | from Freeway Management Systems - USDOT, available at www.its.dot.gov/tcomm/itibeedoc/fms.htm |
| National - Crash reduced by 24% to 50% | from Transportation Planning and ITS: Putting the Pieces Together - FHWA, 1998 |
| National - Crash reduced by 15% to 50% | from ITS Benefits: 2001 Update - Mitretek and Ramp Metering: A Review of the Literature - Arnold |
| California - Crash reduced by 20% to 50% | from Ramp Metering Status in North America: 1995 Update – FHWA |
| Minneapolis/St. Paul - Crashes increase by 26% in peak period when meters were turned off | from Twin Cities Ramp Meter Evaluation, prepared for Minnesota Department of Transportation by Cambridge Systematics, February 2001 |
Risks to Implementation
Risks and factors from a technical implementation standpoint include:
- Difficulties in obtaining reliable data to perform meaningful crash rate and conflict analysis
Risks and factors from an institutional implementation standpoint include:
- Better coordination among agencies responsible for sending vehicles and equipment for incident management
- Interagency coordination to provide more rapid dissemination of accurate and timely traveler information
- Local cities and Caltrans failing to agree on a mutually acceptable ramp metering strategy
A.5 ICM Factors Contributing to Mode Shift
In addition to improving delay, reliability of travel time, and safety proposed ICM strategies are expected to also improve mode shift on the I-880 corridor. The proposed ICM strategies are estimated to improve mode shift as follows:
- Additional mode shift of 2-4% of total corridor trips in response to traveler information about recurrent and non-recurrent congestion
The following table summarizes observed improvements in mode shift, throughout the country. Sources of this information are also provided in the table.
| Expected mode shift | Source |
|---|---|
San Francisco Bay Area (survey):
|
from Traveler Response to Traffic Information on an Incident: A Case Study of the US-101 Corridor in the San Francisco Bay Area - Ronald Koo, Harvard University and Youngbin Yim, University of California at Berkeley |
Seattle (Traffic Reporter) (PC-based graphical, interactive, real-time, traveler info system):
|
from "Traffic Reporter: A Real-Time Commuter Information System" in Applications of Advanced Technologies in Transportation Engineering, Proceedings of the Second International Conference, August 1991 - Edited by Y. Stephanedes and K. Sinha |
| Seattle/Boston (surveys) - When provided with better traveler information, 5% to 10% change travel mode | from ITS Benefits: Continuing Successes and Operational Test Results – Mitretek |
San Francisco Bay Area (TravInfo)
|
from TravInfo Evaluation: A Study of Transit Information Callers - Youngbin Yim, Ronald Koo and Jean-Luc Ygnace |
| Atlanta - About 65% of all survey respondents said they altered the routing, timing, destination, or mode as a result of the device | from Atlanta Traveler Information Showcase 1996 Fact Sheets, October 1996 - Walcoff & Associates, Inc., October 1996 |
Risks to Implementation
Risks and factors from a technical implementation standpoint include:
- Upgrade existing 511 system to improve user access especially to real-time information about service interruptions and connection problems of transit
- Integration of multiple existing systems (of all modes) into a unified system that can utilize all traffic information and disseminate the information to public and transit agencies in a timely manner
- Development of reliable models to predict near-future conditions as a result of mode shift
- Development of reliable methods to automatically generate alternative transit route and transfer locations based on real-time transit conditions
Risks and factors from an institutional implementation standpoint include:
- Transit agencies may need to consider appropriate access control schemes for available parking to control demand during normal operations, including parking pricing and reservation systems
- There is a need for updating operations guidelines to incorporate enhanced interagency coordination strategy for information delivery
- The proposed ICM strategy is more complex than the current approach; more training of traffic and transit operations personnel for response planning will be needed
- Transit agencies should be flexible enough for real-time adjustments in their capacity to become adaptive to problems or changes in other systems