Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration

Office of Planning, Environment, & Realty (HEP)
Planning · Environment · Real Estate

HEP Events Guidance Publications Glossary Awards Contacts

Florida Department of Transportation (FDOT) Express Toll Lane Modeling Workshop Peer Review Report

3.0 FDOT's Current Managed Lane Modeling Practice

The morning session featured an introduction to the FDOT overall process for project development followed by presentations discussing FDOT's experience in evaluating managed lanes. Presentation topics included: a briefing on Florida project development and evaluation processes, the evolution of the FSUTMS toll modeling application in the three phases of I-95 Express Lane development; approaches to traffic operational analysis; approaches to Traffic and Revenue (T&R) studies; and a briefing on risk analysis strategies applied in T&R studies. Together, these presentations provided a context for discussions later in the workshop regarding best practices and steps for moving forward.

3.1 Project Development Process and Evaluation

Florida follows the federal planning process, which begins with the Metropolitan Planning Organization's (MPO's) preparing a long-range transportation plan (LRTP) following the 3-C planning process. Each LRTP covers a span of 20 years. In Florida the LRTP list of needed improvement projects is prioritized considering benefits and cost feasibility. The LRTP improvements list is used by the MPO and FDOT to develop the Transportation Improvement Program (TIP), which consists of a five-year program of projects of which one year is current and four are proposed. Each year, the TIP is modified by adding a new fifth year and advancing the first of its future years to current status. 

As stated in the MPO Program Management Handbook (2012), "(each) MPO carries out three major work activities:

  1. The development and maintenance of the LRTP which addresses no less than a 20-year planning horizon.
  2. The update and approval of the Transportation Improvement Program (TIP), a five-year program for highway and transit improvements.
  3. The development and adoption of the Unified Planning Work Program (UPWP)"

The FDOT has also established an Efficient Transportation Decision Making (ETDM) process that includes planning, programming, development, and environmental phases of project implementation. These phases of project implementation are illustrated in the Figure 3.

Figure 3: FDOT ETMD Process

Figure 3 provides a flow chart describing FDOT's Efficient Transportation Decision Making process, which connects planning, programming, and project development and engineering phases to various processes including community coordination and Environmental Technical Advisory Team Coordination.

These traditional activities were placed in a matrix, shown in Table 2, and presented by Hugh Miller during the workshop's introductory presentation. The issues involved in Florida's project development process are well known by the FDOT staff and supporting consultants. Product deliverables have specific names, and milestones are well established for completing the projects. For traditional roadway projects, the demand models used for project forecasts and traffic operational analysis procedures have been well-established and are updated regularly. In adding express lane projects, FDOT Districts have dealt with more complex demand modeling procedures and more complex traffic analysis procedures (like microsimulation), as well as financial feasibility calculations that require T&R studies.

Table 2: FDOT Project Development Process

 

Traditional Activities

 

Name

Issues

Products

TDM Model

MPO

Regional Planning Studies

Needs & Cost Feasible

LRTP/TIP/UPWP

MPO Model Update

FDOT

Project Planning Studies

Project need and purpose; funding sources and priority;

Concept Report with tentative project features and planning level cost estimates

Latest version of MPO Model

 

Project Development and Environmental Studies

Environmental feasibility; public acceptance; project location, funding and design features

EIS; permits; refined design with cost estimate and schedule

Project validated version of the MPO model

 

Design Studies

Detailed design features

Design Package, R/W

Same model

3.2 Use of SERPM for I-95 Express Lanes

The travel demand modeling conducted to date for express lane projects has been completed as part of the FDOT project development process. Each project makes enhancements to the latest version of the MPO regional travel demand model. The decision of what enhancements to make has been left to the discretion of the consultant performing the work with input from the District modeling staff.

In the case of the I-95 Express Lanes Study, Ken Kaltenbach described the evolution of modeling approaches used as the project advanced from the initial Phase I to the Phase II and Phase III extensions of the express lanes. During Phase I, an analysis of the demand for the I-95 Express was performed using the traditional CTOLL parameter within the standard FSUTMS assignment process. This was done due to severe time constraints, and it used a dynamic toll function based on the express lane volume-to-capacity value to determine the toll.

In Phase II, the I-95 Corridor Planning Study, a binary logit choice model for tolled route/non-tolled-route choice was implemented during highway assignment. This was combined with more extensive feedback and the use of a subarea model to reduce the network size. This study examined the feasibility of high-occupancy vehicle (HOV) and high-occupancy toll (HOT) lanes for 63 miles between Griffin Road and Indiantown Road.

Finally, in Phase III's I-95 Project Development and Environment (PD&E) Study, an enhanced version of the previous study's model was used, retaining the binary choice model implemented in the assignment phase but refining the model parameters to increase sensitivity to changes in tolls and travel times.

Figure 4 illustrates the three phases of the I-95 Express Analysis.

Figure 4: I-95 Express Analysis Flow Chart

Figure 4 provides a flow chart detailing the interrelated processes involved in the three phases of the I-95 Express Analysis. The figure addresses the forecasting, operations, and managed lane design analyses over the time of each phase.
Source: I-95 Corridor Planning Study: Managed Lane Feasibility, FDOT District 4, July 2012

3.3 I-95 Express Operational Analysis

With respect to traffic analysis for I-95 Express Phase I and II, David Stroud described how traffic analysis based on Highway Capacity Manual Software (HCS) in the PD&E and Interchange Operational Analysis Report (IOAR) did not provide a sufficient assessment of the complex traffic weaving maneuvers associated with managed lanes. He explained that the selection of a microsimulation approach (using CORSIM) as the traffic analysis tool was based on the need to assess transportation system complexities associated with adding a system of tolled managed lanes to the currently congested and geometrically constrained interstate system.

A team of technical experts from FHWA Florida Division, FDOT Central Office, and FDOT District Office met regularly to discuss the application of the FHWA Traffic Analysis Toolbox (TAT) Volume IV microsimulation guidelines and procedures to the project and to resolve technical issues. Key challenges addressed during these meetings included: the proper selection of temporal and spatial limits; accurate and consistent coding of 24 centerline-miles of freeway, 18 miles of service roads, and four system-to-system interchanges; gathering input data from available sources; and calibrating the model to produce outputs that were verifiable and reproducible. Speed, volume, density, and queuing were the primary measures of effectiveness (MOEs) for model calibration. The microsimulation model's operational analysis resulted in design changes that improved safety and operations.

For the I-95 Express Phase III traffic analysis, both planning and PD&E studies were conducted. The I-95 Corridor Planning Study (CPS) included a rigorous process to develop a methodology to integrate travel demand, traffic operations, and managed lane design into the project development process. The methodology includes a multi-resolution analysis of freeway traffic operations, where Highway Capacity Manual (HCM) tools are used in the initial planning stages and microsimulation is used during the conceptual development or PD&E stage of the project development process. The intent is to focus the planning study efforts on verifying the need to implement tolled managed lanes to reduce congestion and improve interstate mainline operations. Once the need is established by the planning study, the PD&E study uses HCM software and microsimulation to assess the complex traffic maneuvers created by the implementation of tolled managed lanes with placement of ingress and egress access points. The CPS verified the need for implementing tolled managed lanes and recommended the location of ingress and egress access points based on multimodal considerations and market demands between interchanges in the study area.

The PD&E for I-95 Express Phase III was divided into three segments with separate PD&E studies pursued simultaneously. A rigorous process for developing the Traffic Methodology Memorandum for tolled managed lanes was undertaken at the beginning of the studies. The memorandum applied to each of the studies, specifying the details of data collection; travel demand modeling and forecasting; and traffic operational analysis methods using HCM software and microsimulation with VISSIM) which followed the guidelines of FHWA TAT Volume III.

Additionally, the memorandum specified the use of diurnal factors to convert the travel demand model peak-period volumes to peak-hour volumes and calibration targets for hourly traffic flows, travel times, and queuing. The pricing and route choice elements of the microsimulation software's managed lane module were applied to the build conditions to determine the traffic demand for the tolled managed lanes. MOEs included volumes and speeds of both the general purpose and managed lanes, network-wide assessment of no-build and build conditions, and managed lane revenues.

3.4 Summary of Managed Lane Program

In a discussion of current Florida modeling practice in relation to tolled facilities, Steve Ruegg described the recent history of toll modeling incorporation in the FSUTMS framework. Since the start of the FSUTMS model design, there has been some accommodation of tolled facilities, with the Turnpike's fixed toll highways in mind. The model structure has used an assignment-based fixed-toll algorithm, with the toll itself being converted to a value-of-time based on the "CTOLL" value. In addition, toll collection facilities are modeled explicitly, with deceleration, toll service time, and acceleration times explicitly modeled through the use of specialized toll links. Service times are estimated based on a multi-server queuing model embedded into the traffic assignment's volume-delay functions. Later improvements have included ramp-to-ramp tolling structures to better represent the Turnpike pricing system.

In the past three years, FDOT Central Office has undertaken two major projects, which will improve the ability to model express lanes in Florida. The first is a major review and upgrade of standard modeling practices, as reflected in the FSUTMS. This improvement program, known as the Transit Model Update (TMU) made several improvements to the four-step model procedure, including the following:

  1. Expanded trip purposes;
  2. Time-of-day stratification;
  3. Use of feedback to trip distribution;
  4. Use of a destination-choice distribution model formulation;
  5. Expanded mode choice nesting structure and mode options; and
  6. Tighter assignment closure criteria.

These model improvements enhance express lane modeling by improving overall model sensitivities and capabilities. For example, time-of-day modeling will allow for more precise estimation of variable tolls and the effect of them on distribution and mode choice.

A second initiative by the FDOT Central Office was specifically directed at improving managed lanes modeling. This program included a "toolbox" of three model approaches that address modeling dynamic open-road tolling in Florida as shown in Table 3.

Table 3: FDOT Managed Lane Modeling Concept Plan

  Phase I Phase II Phase III
Type Assignment-Based Integrated Toll Choice Mode Choice + Assignment Discrete Choice
Model Type Trip-Based, Static Trip-Based, Static AB and DTA
Features Dynamic toll Estimation, WIllingness to pay Curve, Toll Policy Feedback of toll LOS skims to mode choice. Sensitive to multi-modal shifts Incorportaes detailed HHLD characteristics for toll choices
Uses LRTP & Corridor Planning Multi-modal corridor evaluation Policy Sensitivity tetsing, and TP Planning
Data Requirements SR/RP survey for WTP curve or logit estimation SP+RP survery to estimation &calibrate MC parameters HIS supportive of AB models
Availability Summer, 2012 2013 2014-2015

 

The development plan consists of three phases, with the intent to generate a robust toolbox of managed lane modeling applications that can meet the planning needs of all agencies based on their modeling capabilities and the required level of detail and model sophistication. In the three-phase program, Phase I consists of developing an assignment-based dynamic toll model, featuring dynamic toll estimation and shift in toll paying demand, willingness to pay curves, and sensitivity to various toll policies. Phase I development work was completed in the summer of 2012.

Phase II of the program extends the toll effects to the mode choice level and focuses on implementing toll choice within the mode choice model. This mode choice model with toll paying alternatives was then integrated with the assignment-based dynamic toll model of Phase I, using a feedback structure. The Phase II model provides the ability to forecast occupancy level shifts and mode shifts, and a more comprehensive representation of the utility of a toll facility than is possible in the static highway assignment model. The development of Phase II prototype models is also complete, and final documentation was published in April of 2013.

Finally, Phase III focuses on implementing managed lanes within an activity-based model (ABM) framework. The intent is to take advantage of the detailed person and household attributes available in an ABM to better represent the factors that affect the choice of managed lanes. Activity-based models have been extensively used to support road pricing projects and should be more appropriate for policy sensitivity testing of managed lanes. Phase III is expected to commence in the summer of 2013.

3.5 Florida Turnpike Experience

Josiah Banet of URS/FTE presented on the Turnpike's experience with modeling express lanes in Florida and identified four types of T&R studies, as shown in Figure 5. From the top to the bottom of the figure, the types of studies imply an increasing level of detail, time, and resources to generate the desired output information. For each type of study, there is a modeling approach suited to provide the needed information for each study.

Figure 5: Types of Traffic & Revenue Studies

Figure 5 provides a depiction of four varieties of traffic and revenue studies, including top down, sketch level, planning level, and investment grade.

In a "Top Down" T&R study, the focus is to collect general design and performance information on the proposed facility, including lanes, traffic volumes (existing and future), and access design. Operational assumptions, such as types of vehicles allowed, will also be defined. The analyst then compares these features to known performance of other, similar facilities to determine initial feasibility or screening for further studies.

A "Sketch Level" T&R Study (Level 1) is used to determine the high-level financial feasibility of an express lane project using limited data sources. The results of a "Sketch Level" study would assist an agency in determining the need/desire to commit additional resources to further assess a project's feasibility as part of an express lane system. This approach to evaluating an express lanes project within a limited-access facility utilizes existing modeling tools that rely on available data. This approach does not include refinement to the traffic forecast models or additional data collection. Unavailable forecast years for traffic are developed based on reasonable assumptions regarding growth rates. Post-model adjustments to forecasts may be performed based on historical traffic data. In some cases, a stand-alone corridor time-of-day model is used to produce toll rates based on the level of congestion throughout the day. The T&R results are corridor-level and considered to be average estimates with a positive or negative deviation, which are refined with more detailed analyses from a "Planning Level" study.

A "Planning Level" T&R Study (Level 2) results in a more detailed revenue forecast for a project that the agency can then use for developing a project funding plan. This "Planning Level" approach would refine the traffic forecast models to better reflect the selected project corridor's traffic patterns with updated roadway network and improved land use assumptions for existing and future years. A minimum of two forecast years would be developed based on the latest socioeconomic data (e.g., population and employment estimates). Refinement to the validation of the traffic models would be undertaken in order to improve model accuracy and provide more confidence in the forecast results. A time-of-day toll diversion model is used to define traffic and toll rates by hour. Additional data collection efforts would be undertaken to ensure the most recent traffic patterns for the project study area are accurately reflected in the traffic models. This could include, but is not limited to, an origin-destination (O/D) study, speed study, or stated preference survey (SPS). Statistical probability analyses are employed to assess project risk. 

An "Investment Grade" study (Level 3) is required when the financing mechanisms use bonds to leverage against anticipated project revenues and the project is ready to go to market. In this case, FDOT has a high confidence that the project will be implemented and would like certainty in the accuracy of the revenue projections. At the "Investment Grade" level, primary data collection of O/D patterns and value of travel time savings (using SPS results) is often required. An independent assessment of the socioeconomic data affecting project traffic is undertaken. The travel demand model is also highly specialized and adapted specifically for the project corridor. Efforts to improve the model forecasting by incorporating reliability are imbedded into the modeling process, and microsimulation models are also employed to provide a more accurate assessment of corridor travel times.

3.6 Quantifying Forecasting Risks

Risk can be defined as the ability of a project to achieve an objective; whereas, uncertainty deals with potential fluctuations in circumstance that may or may not yield significant impact on the project's ability to reach an objective. Tom Adler of RSG first discussed methods of identifying and quantifying the uncertainties and risks associated with model forecasts and then presented an example from work conducted for the Orlando I-4 T&R Study managed lane project.

Dr. Adler began by identifying three main sources of uncertainty associated with travel forecast models that will subsequently impact project risk, including:

  1. Model Structure and Data;
  2. Analysis Bias; and
  3. Inherent Uncertainties about the Future.

The objective of his presentation was to describe methods to quantify these inherent uncertainties within models. While it is recognized that computational methods, statistical methods, and qualitative analysis of uncertainty in models might be used to produce measures of uncertainties from model execution directly, practical problems, including run times, may prevent this approach.

Dr. Adler stated that corridor traffic alternative results generated by a series of travel demand model runs may be subsequently represented using a multivariate closed-form function of model inputs. Application of this type of function allows for a rapid calculation of alternatives represented by variations in key inputs and can therefore generate a response surface which can be used to evaluate the probability of risk associated with the probability uncertainties in input value ranges.

In the Orlando I-4 T&R Study, illustrated in Figure 6, a 21-mile managed lane project, adding two priced lanes in each direction was evaluated using this approach. Nine "experimental" sensitivity tests were performed in which travel times, toll rates, economic forecasts, and network structures were varied. From these results, a synthesized model for T&R estimation was developed that was able to match model results with an r-squared value of 0.98. Using this synthesized model, probability distributions of revenue and traffic were generated, representing one million random draws from input distributions.

The presentation concluded that response surface models can be developed with accuracy and can be used to effectively simulate risks associated even with more complex travel demand models.

Figure 6: I-4 Managed Lanes Corridor Map

Figure 6 provides a map that highlights the I-4 Managed Lanes Corridor study area and surrounding related facilities in Seminole and Orange Counties.

Updated: 03/28/2014
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000