12th International HOV Systems Conference: Improving Mobility and Accessibility with Managed Lanes, Pricing, and BRT

Conference Proceedings

BREAKOUT SESSION — PLANNING/MODELING FOR MULTIPLE USER GROUPS

Managed Lanes Modeling Process

Gustavo Baez, Wilber Smith Associates

Gustavo Baez discussed the modeling process for managed lanes. He provided a definition of managed lanes and presented the goals of the modeling process. He described the types of modeling studies typically conducted in Texas and illustrated issues that may be encountered in the modeling process.

A managed-lane facility is defined as one that increases freeway efficiency by packaging various operational freeway efficiency and design actions. Managed lanes operations may be adjusted at any time to better match regional goals. The goal of the modeling process is to provide a set of tools to allow decision makers to select operational alternatives which optimize capacity, speed and/or speed, and revenue.
Three types of modeling studies are typically conducted in Texas. A Level 1 or sketch-level evaluation takes approximately two-to-four weeks to conduct. A limited amount of information is needed for a sketch-level analysis, which provides a gross estimation of use levels and revenues.
A Level 2 or preliminary study involves a greater level of detail. Level 2 studies take six-to-eight months to conduct. Data needed for a preliminary evaluation includes the existing and future highway configuration, speed and delay, traffic counts, and demographic and trip tables provided by the local metropolitan planning organization.
The Level 3 or investment grade study is the most detailed evaluation, typically taking 10-to-12 months to conduct. Data needed for a Level 3 evaluation includes the existing and future highway configuration, speed and delay, traffic counts, and demographic and trip tables. In addition, an independent review of the demographic data is conducted. Stated preference surveys are also conducted to obtain a value of time distribution.
A number of factors are important to consider in forecasting demand for a toll road. Examples of these factors include competing facilities, length of the facility, the level of congestion, and travel time savings. Other factors to consider include forecasted growth, the value of time, and willingness to pay. The inclusion or exclusion of a truck component will also need to be examined.
A number of factors are also important to consider in forecasting demand for a managed lane. These factors are similar to those considered with a toll road. A significant difference with most managed-lane projects is that they are adjacent to or part of a freeway. As a result, for a managed-lane project, the freeway must be very congested and must be failing. An essential element of managed-lane projects is providing a travel time savings over the general-purpose freeway lanes. Time of day modeling is critical with managed lanes; especially examining the shoulders of the peak period versus the peak hour.
Three levels can be studied when examining global demand. The maximum assumes the HOV or managed lane is free and available to anyone. The optimum is based on a toll rate and HOV class allows toll-free. The minimum assumes the HOV and managed lanes is very expensive for SOVs and toll-free only for HOV drivers. The global demand is not static, however, and will be influenced by the time-of-day and the direction of travel.
The travel demand model can be run with different trip tables. The traffic simulation model estimate traffic operations. The models should be run for different time periods, the shoulders of the peak periods, and by directionality.
Other elements may need to be considered in the modeling process. Short trips versus long trips and the minimum toll and the maximum toll should be considered. The current and projected transit use in the HOV lane should be explored. The toll rate variation and the use of a flat toll rate or a dynamic toll rate represent other considerations.

Estimating the Benefits from BRT/Managed Lane Alternatives Using SMITE-ML 2.0

Patrick DeCorla-Souza, Federal Highway Administration

Patrick DeCorla-Souza discussed the use of SMITE-ML to estimate the potential benefits of BRT and managed lane alternatives. He described the purpose of sketch planning, the features of SMITE-ML, and the use of the model. He also presented a case study application.

Sketch planning can be used with BRT and managed lanes for a number of purposes. First, it can be used to compare managed lane concepts in terms of performance, social benefits, financial feasibility, and revenue cost. Sketch planning provides a quick response and typically needs only minimal data. Sketch planning provides a screening-level evaluation. It also provides an understanding of the trade-offs associated with various project components.
SMITE-ML can be used with a variety of alternatives. These alternatives include conventional approaches, such as free general-purpose lanes and HOV lanes. It can also be used with pricing options, including HOT lanes, fast and interwined regular (FAIR) lanes, and FAIR highways.
SMITE-ML requires a number of inputs. The first inputs are daily traffic volumes and hourly capacities for the freeway and arterials. The base mode shares and travel time and cost changes are other inputs. The travel time and cost changes can be checked against output time and toll rates.
The first step uses the no-build forecasts and the time and cost changes as input values into the Pivot Point Logit Model. The output from this step is an estimate of vehicle trips by mode. In the second step vehicle trips by mode and road capacities are input into the equilibration to estimate induced and diverted traffic. If a toll lane is being considered, a toll factor would be added in this step. In the third step, traffic estimates and road capacities are input into the impacts estimation element. The results of this step are estimates of speeds, delays, revenues, user benefits, and social costs.
SMITE-ML provides a number of outputs. These outputs include estimates of travel demand, speeds, and delays. Other outputs include estimates of toll revenues and external costs. SMITE-ML also provides an estimate of performance in terms of economic efficiency and cost-effectiveness.
The case study application included five alternatives — 10 general-purpose lanes, four express toll lanes and six general-purpose lanes and moderate transit, four express toll lanes and six general-purpose lanes and BRT, four HOT lanes and six general-purpose lanes with moderate transit, and four HOT lanes and six general-purpose lanes with BRT. The model provides an estimate of the change in HOV and daily transit person trips, hours of delay, toll revenues versus costs, and present value of benefits versus costs for the different alternatives.
The results of the case study application indicate that express toll scenarios increase revenues and financial feasibility. The HOT lane scenarios increase benefits and net present value, but reduce revenues. The alternative with BRT increase benefits and net present value but reduces financial feasibility. The result of an additional sensitivity analysis indicates that demand elasticity has a small effect on revenue and benefits, and the value of time has a large effect on revenue and benefits.
SMITE-ML can provide quick response estimates of impacts for pricing policies. SMITE-ML is available at www.fhwa.dot.gov/steam — go to related links.

Management of Special Use Lanes: SUL Model Development and Analysis

Yuko Nakanishi, Nakanishi Research and Consulting

Yuko Nakanishi discussed the application of simulation and mathematical models with special use lanes. She described some of the benefits and issues associated with the use of simulation and mathematical models, as well as the methodology for applying them with special use lanes.

There is an irony associated with special use lanes. During ideal conditions, the demand for special lanes is low. During congested conditions, however, special use lane demand is high, but may be difficult to access. Thus, a problem may be that while special use lane capacity is available, drivers who want to use it cannot. This means that the expected benefits of special use lanes will not occur.
A potential problem with HOV facilities relates to the cost of constructing the lanes. The cost of constructing HOV lanes on an existing right-of-way may range from a low of $30,000 per lane mile to a high of $2 million per lane mile depending on a variety of circumstances. Construction costs for HOV lanes in an exclusive right-of-way may be as much as $25 million per lane mile.
Models can be used to help assess different special use lane alternatives, origin and destination patterns, hot spots, and transfer lanes. Models can estimate the demand for various strategies.
Examples of simulation models may include microscopic models, mesoscopic models, and macroscopic models. Examples of microscopic models include CORSIM and WATSIM. TRANSIMS is an example of a mesoscopic model and Freq12 is an example of a macroscopic model.
Mathematical models include the attributes of ability to estimate capacity and the ability to incorporate demand scenarios. Extensive field observations are not needed with mathematical models. It would also not be feasible to collect the data at the necessary level of detail.
There are potential issues that should be considered with the use of mathematical models. Examples of these issues include the need to consider vehicle type, driver characteristics, and gap acceptance. The visualization capability of mathematical models may also be limited.
In a methodology flow chart, selecting objective functions and establishing major constraint sets feed into the process to estimate dimensionality, which feeds into formulating the equation set. The next steps are to select user friendly software and then to run the model and display the results.
A number of key issues should be examined in applying models with special use lanes. It is first important to define the objective and perspective for a project. This step may be considered as defining the project objective function. A second issue is to identify the highway segment, including how long it is and how many lanes it has.
There are more mundane issues associated with computer capacities and budget. The size of the segment, and speed, memory, and spreadsheet capabilities of a microcomputer may limit the models that can be used and the analysis techniques. Issues to consider include processing time per run, the availability of computers, and researchers available to perform the runs. The amount of available funding should also be considered. The desired graphical output and the potential need for programming staff may also be issues.
While mathematical models do not simulate individual vehicles and track their movements over time as microscopic models do, they provide the ability to incorporate elements such as driver demand characteristics and gap acceptance parameters. Graphics of capacity utilization and visualizations of hot spots are possible. Also, visualization of the impacts of origin and destination patterns and conflicting flow patterns is possible. When combined with actual geometric data, the mathematical models can be a useful planning tool in determining the effects of different origin and destination patterns. This approach is possible without the significant data requirements by other models.

Coding BRT, Park-and-Ride Lots and Transit in the Context of a Dynamic and Interactive Regional Traffic Model at the Atlanta Regional Commission

Guy Rousseau, Atlanta Regional Commission

Guy Rousseau discussed the process of coding BRT, park-and-ride lots, and public transportation services in the regional traffic model at the Atlanta Regional Commission (ARC). He described the regional traffic model, the definitions and coding requirements of different transit elements, and some of the issues encountered with use of the model.

The definition of the Atlanta region varies. There are 10 counties included in the planning area. However, 19 counties contain a portion of the 2000 Atlanta urban area boundary. The ARC expands to 20 counties. A total of 13 counties were classified as serious ozone nonattainment areas by the Clean Air Act Amendments of 1990. These counties were reclassified to severe nonattainment in January 2004, and 20 counties are included in the eight-hour nonattainment area.
Accommodating the projected growth in population represents a major challenge for the region. The forecast of 2.3 million by 2030 people is equivalent to adding the metro area of a city the size of Portland, or two cities the size of Jacksonville, or four cities the size of Chattanooga. The region's employment level in 2030 is estimated at 1.7 million.
Three different types of freeway interchange are coded in the system. These interchange types are freeway to local, three-legged freeway-to-freeway, and a legged freeway-to-freeway. All ramps are coded with two lanes for all interim interchanges. The capacity of a freeway to local interchanges is 1,200 vehicles per lane with a speed of 25 mph. The capacity of a freeway-to-freeway interchange of 1,800 vehicles per lane with a speed of 40 mph.
There are five transit files. These files are premium transit, non-premium transit, rail network, premium park-ride lot, and non premium park-ride lot. The premium transit file includes very reliable service, prepaid boardings, and MARTA rail type stations. The non-premium transit file includes buses operating in traffic where reliability of service is dependent on highway congestion.
The rail network file is used for coding premium transit links that are not part of the highway network. The text file contains link attributes for mode, station-to-station distances, speeds, and directionality. The premium park-and-ride lot file allows for drive times of up to 60 minutes. The non-premium park-and-ride lots file provides for shorter drive times of up to 15 minutes.
Local bus service is provided at stops along the majority of the routes and operates primarily on arterials, collectors, and local roads. Express buses provide limited stops along routes and generally operate from bus stations or park-ride lots with alignments along interstates and freeways. The coding requirements include peak and off-peak headways and route alignment for non-premium transit. Bus speeds are calculated based on the highway speeds during the model run. For local bus service, the stops should be coded so that all zones along route have walk access to transit unless otherwise specified. For express bus service, stops should only be coded at specified locations, such as park-and-ride lots or bus stations
The fixed-guideway file is defined as transit operating in its own right-of-way with permanent boarding station, such as the MARTA rail system. Coding requirements include station locations, station-to-station distances, station-to-station speeds, and peak and off-peak headways. Coding rail stations involves connecting rail lines to station nodes and routing buses to the feeder bus node to allow people to transfer at the rail station.
Three types of BRT facilities are included. The first type of BRT in a busway operated in bus-only lanes at fixed speeds with stations and pre-paid boardings. The second type of BRT in HOV lanes operates with buses in the HOV lanes with on-line stations and pre-paid boardings. The third type, arterial BRT, operates on arterials with queue jumping lanes, signal preemption, and pre-paid boardings. The requirement coding information includes station locations, route alignments, BRT stop patterns, BRT and feeder bus headways, and station-to-station run speeds and distances.
Examples of coding the different networks highlight how the process is used. Ensuring that each alternative is coded correctly is critical to valid models and model outputs.

Back to Top