4.0 Test Corridor Data
The I–880 test corridor, shown in Figure 4.1, has been studied by many transportation agencies, including Caltrans, the ACCMA, and the California Center for Innovative Transportation (CCIT). This section provides a summary of available test corridor data organized in six subsections, including macroscopic travel demand model data, mesoscopic simulation model data, microscopic simulation model data, zone correspondence data, traffic data, and transit data. The electronic data files are also provided in the data DVD.
The test corridor was modeled in CUBE, Dynasmart–P, and Paramics macroscopic, mesoscopic, and microscopic models, respectively. The validation and calibration traffic data are adequate for freeways, but quite limited for arterials and local streets.
4.1 Test Corridor AMS Framework
The approach adopted for the test corridor analysis applies the framework from the AMS Methodology document, as shown in Figure 4.2. The AMS methodology for Test Corridor applies macroscopic trip table manipulation for the determination of overall trip patterns, mesoscopic analysis of the impact of driver behavior in reaction to ICM strategies (both within and between modes), and microscopic analysis of the impact of traffic control strategies at roadway junctions (such as arterial intersections or freeway interchanges). The methodology also includes a simple pivot-point mode shift model and a transit travel time estimation module, the development of interfaces between different tools, and the development of a performance measurement/benefit-cost module.
4.2 AMS Tools
The Test Corridor modeling approach encompasses tools with different traffic analysis resolutions. All three classes of simulation modeling approaches – macroscopic, mesoscopic, and microscopic – will be applied for evaluating ICM strategies. The objective of the modeling approach is to provide the greatest degree of flexibility and robustness in supporting subsequent tasks for the Test Corridor and AMS support of Pioneer Sites. This section describes the various off-the-shelf and custom tools applied for the Test Corridor to conduct the modeling of the ICM strategies.
Travel Demand Forecasting Model
Predicting travel demand requires specific analytical capabilities, such as the consideration of destination choice, mode choice, time-of-day travel choice, and route choice, as well as the representation of traffic flow in the highway network. These attributes are found in the structure and orientation of travel demand models; these are mathematical models that forecast future travel demand from current conditions, and future projections of household and employment characteristics.
A calibrated CUBE travel demand model (TDM) of the Alameda County, shown in Figure 4.3, will be used to develop the trip tables for the Test Corridor. Two levels of subarea trip tables will be developed from the TDM – to cover the spatial extents of the mesoscopic and microscopic simulation models. The travel demand model also will be used as the analysis engine for a simple pivot-point mode-choice model, which will analyze the mode shifts due to ICM strategies. The output from mode choice analysis and trip table manipulation will be static corridor-based trip tables that take into account basic trip impacts associated with corridor conditions, current operations, or operational changes. A detailed description of the mode choice model is provided later in this section.
Figure 4.2 Test Corridor AMS Framework
Figure 4.3 Alameda County Travel Demand Model
Mesoscopic Simulation Models
Mesoscopic models combine properties of both microscopic and macroscopic simulation models. The mesoscopic models’ unit of traffic flow is the individual vehicle, and they assign vehicle types and driver behavior, as well as their relationships with the roadway characteristics. Their movement, however, follows the approach of macroscopic models and is governed by the average speed on the travel link. Mesoscopic model travel prediction takes place at an aggregate level, and does not consider dynamic speed/volume relationships as reflected in queue lengths and the temporal distribution of congestion. As such, mesoscopic models provide less fidelity than microsimulation tools, but are superior to travel demand models, in that, mesoscopic models can evaluate dynamic traveler diversions in large-scale networks.
A DynaSmart-P mesoscopic model of the subarea, which extends beyond the mainline I–880 corridor, will be used for the analysis of ICM strategies of the Test Corridor. The DynaSmart-P network will use a trip table from the travel demand model. After the subarea network is extracted from the macroscopic travel demand model, the subtracted network will then be converted into a dynamic network in the mesoscopic simulation model. Note that DynaSmart-P does not use centroids and centroid connectors; rather, it directly generates vehicles on generation links in a zone. Also, zonal aggregation will be needed to construct the dynamic network.
The standard bi-level origin-destination (O–D) estimation approach will be used. The upper problem is a variation of General Least Square (GLS) of the difference between simulated and observed volumes, and the lower problem is the DynaSmart–P dynamic traffic assignment. The module starts with the seed O–D demand matrices from the static network. The outputs from this module are dynamic O–D demand matrices. The matrices reflect paths and departure times as in the meso-model.
The model will be used to support the analysis of the dynamic impact of ICM strategies that try and induce shifts of trips from one network to another, such as pricing, and corridor-specific traveler information (pre- and during trip). An illustration of the mesoscopic network is shown in Figure 4.4. The network characteristics are presented in Table 4.1.
Figure 4.4 Mesoscopic Simulation Network for the Test Corridor
Table 4.1 Mesoscopic Network Characteristics
Network Data |
Value |
---|---|
Number of nodes |
2,658 |
Number of links |
6,888 |
Number of zones |
1,078 |
Microscopic Simulation Models
Microscopic simulation models simulate the movement of individual vehicles, based on theories of car-following and lane-changing. Typically, vehicles enter a transportation network using a statistical distribution of arrivals (a stochastic process); and are tracked through the network over small time intervals (e.g., one second or fraction of a second.) Typically, upon entry, each vehicle is assigned a destination, a vehicle type, and a driver type. In many microscopic simulation models, the traffic operational characteristics of each vehicle are influenced by vertical grade, horizontal curvature, and superelevation, based on relationships developed in prior research. The primary means of calibrating and validating microscopic simulation models is through the adjustment of driver sensitivity factors.
A Paramics microsimulation model for the Test Corridor is currently being developed for other studies. Depending on the delivery schedule for these other studies the Paramics model can be used to support the evaluation of the operational control aspects of the ICM strategies, such as ramp metering strategies. Microscopic simulation analysis will output detailed travel times that can be used to augment the mesoscopic simulation analysis. This augmentation entails the conversion of operational impacts identified at the microscopic level into adjustment factors at the mesoscopic level. These factors can support the modification of the mesoscopic analysis, such that the impacts of the operational control aspects of ICM strategies can be analyzed in conjunction with the trip management/shifting aspects of those strategies. An illustration of the Paramics model for the Test Corridor is shown in Figure 4.5.
4.3 Zone Correspondence
Traffic analysis zone correspondence between the macro, meso, and micro models will be created to achieve data transfer between these models. Zone correspondence between the CUBE travel demand model and the DynaSmart-P mesosimulation model will be achieved by converting the subarea model in CUBE to a shapefile, and importing the shapefile into DynaSmart–P.
To create zone correspondence between the meso model and the micro model, the “generating links” feature of DynaSmart-P will be used to create trip tables for the zones corresponding to the Paramics network.
Figure 4.5 Microscopic Simulation Network for the Test Corridor
4.4 Traffic Data
Data for model calibration include time-dependent freeway traffic counts, time-dependent section travel times (based on probe vehicles), intersection turning movement counts, and arterial traffic counts. Data processing will be necessary due to data missing and data inconsistencies. Detail explanations of data processing for ramp and mainline data are provided in this section.
The mainline counts were collected on March 1, 2, and 8, 2005 through PeMS. The averages of counts from these three days are used as the observed mainline traffic counts. Ramp count data comes from two different sources: the TMC and the census data from Caltrans.
Using the detectors installed in the field, the TMC can collect ramp and mainline data everyday with 30-second format. It is the best data source used for model calibration. However, the TMC data suffer from missing data due to detector and communication failures. Currently, PeMS can fill in missing records using historical data of the same detector or neighboring detectors.
4.5 Transit Data
Alameda County Bus Transit Routes
AC Transit operates two major local bus lines (82 and 82L) along I–880 and about 15 express bus lines. Line 82/82L operates 24 hours a day from the Hayward BART station (Bay Fair BART for 82L) to downtown Oakland via East 14th Street and International Boulevard. Figure 4.6 shows the route map. The express lines using I–880 include Line S (South Hayward to San Francisco), Line SA (San Lorenzo to San Francisco), Line SB (Newark to San Francisco), Line OX (Harbor Bay/Alameda to San Francisco), Line O (Alameda to San Francisco), and Line W (West Alameda to San Francisco). Table 4.2 provides a summary of transit service along International Boulevard.
Figure 4.6 Test Corridor Bus Transit System
Table 4.2 Existing Transit Service on International Boulevard/East 14th Street
Route |
Weekday Service – |
Weekday Service – |
Weekday Service – |
Weekday Service – |
Weekend Service – |
Weekend Service – |
Weekend Service – |
---|---|---|---|---|---|---|---|
82 International (downtown Oakland to SL BART) |
24 hrs |
12 minute |
15 minutes |
No service |
24 hours |
15-60 minutes |
No service |
82 International (SL BART to Bayfair BART) |
7:30 p.m.-7:00 a.m. |
No service |
No service |
15-60 minutes |
7:00 p.m.-10:00 a.m. |
No service |
15-60 minutes |
82L International Limited (downtown Oakland to Hayward BART) |
7:00 a.m.-7:00 p.m. |
12 minutes |
15 minutes |
No service |
10:00 a.m.-7:00 p.m. |
15 minutes |
No service |
AC Transit is in the process of implementing Bus Rapid Transit (BRT) between Berkeley and San Leandro along the International/East 14th Street corridor. Phase One will begin in fall 2006 with the initiation of Rapid Bus service (Line 1R) featuring signal coordination and priority, stop amenities, and real-time traveler information. Construction for Phase Two is scheduled to begin in 2008 and will feature dedicated transit ways at a large percentage of its runways and significant ITS and other technological improvements.
Passenger boardings and passenger miles for AC Transit Routes 82 and 82L are displayed in Table 4.3. AC Transit has several major transfer points along the corridor: Fruitvale BART, Coliseum BART, San Leandro BART, and Bayfair BART. Each of these BART stations serves between 5 and 8 bus routes and provides intermodal transfers with the BART service. Over 7,000 passengers per day access BART or buses at these stations. Figure 4.7 shows an example of bus routes accessing BART station at 12th Street, downtown Oakland.
Table 4.3 Existing Bus Operations
Route 82/82L |
Daily Passenger Boardings |
Average Daily Passenger Miles Per Trip |
---|---|---|
Weekday |
16,727 |
244.3 |
Saturday |
10,169 |
139.2 |
Sunday |
9,723 |
173.7 |
Figure 4.7 Bus Routes Accessing BART Station at 12th Street
Bay Area Rapid Transit (BART) Rail
San Francisco Bay Area Rapid Transit District (BART) is a public rail rapid transit system that serves major parts of the San Francisco Bay Area. The total system comprises 104 miles of double track and 43 stations, as shown in Figure 4.8. The BART system along the test corridor includes 20 miles of double track and 12 BART stations. BART is connected to regional rail and bus services.
The combined daily ridership for the A–Line and L–Line, and downtown Oakland stations is close to 100,000, or 25 percent of the total BART’s daily ridership. This ridership includes: approximately 48,000 on the A-Line (Lake Merritt Station to Fremont) or approximately 14.2 percent; approximately 10,000 on the L–Line (Castro Valley and Dublin/Pleasanton stations) or approximately 3 percent; and 29,000 entries or 8.6 percent) for the downtown Oakland stations (12th Street and 19th Street) or approximately 8.6 percent.