Model Development

Below is a list of the software tools used by the modeling staff at DVRPC to estimate the impacts of future projects and current trends on the region.

Travel Demand Models

Travel demand models are used to estimate the outcome of proposed projects or practices. Whether it is an operational change (e.g. – installing dynamic message signs to provide travelers with real-time information or rerouting a bus line) or an infrastructure project such as building a new road or rail line, these models allow planners to anticipate the resulting changes in travel behavior.

TIM 1.0

The Travel Improvement Model Version 1.0 (TIM 1.0) is a traditional four-step travel model, and the current platform used for projects and analysis. TIM 1.0 was based on DVRPC's previous system of 2068 traffic analysis zones. Many model components of TIM 1.0 were unchanged from the legacy TRANPLAN model, in particular trip generation and trip distribution. Other model components, including mode choice, highway assignment, and transit assignment went through significant changes and upgrades, which was necessary to accommodate the model under the new software platform. Major benefits from the migration to VISUM include better graphics and mapping, automated QA/QC, and convergence of the highway travel times which is beneficial in the comparison of network scenarios. TIM 1.0 was used for all new transportation projects that started between April 2010 and July 2013. TIM 1.0 is a traditional four-step travel demand model.

In general, the four-step travel demand model proceeds in this way:

  • The first of the four steps, trip generation, estimates the number of trips produced within specific areas - called traffic analysis zones (TAZs) - based on the population and employment characteristics of that area. Trips are generated for a number of different trip purposes.
  • Trip distribution, which occurs next, determines where the trips produced in the first step will travel to by looking at the ease of getting to every possible location and draw of that area for that trip purpose. This establishes an origin and destination for each trip.
  • The next step, mode choice, determines how a particular trip will be made, such as by driving a car, carpooling, riding transit, or walking.
  • Finally, the model determines which paths along the road or transit network each trip will take to reach its final destination. This step is called trip assignment. Paths are calculated to find those that minimize travel time and cost.

TIM 1.0 documentation may be found here [3.4 MB .pdf] and here.

TIM 2.0/2.1

TIM 2 is DVRPC's current travel demand model, useful for many planning and design applications – highway alternatives analysis, transit alternatives analysis, air quality conformity analysis, long range strategic planning, facility design, economic analysis, greenhouse gas planning, and many others. TIM 2 is a best-in-class 4-step travel demand model. Architecturally, TIM 2 runs via a combination of Python scripts and the VISUM transportation planning software package by PTV AG. TIM 2 was extensively validated against numerous data sources to ensure that it accurately represents current travel behavior in the DVRPC region. DVRPC staff made sure to go beyond a simple “make the numbers match” exercise when calibrating and validating TIM 2. Instead, the emphasis was on ensuring that the TIM 2 model understood travel behavior – the various choices that are made on a daily basis by millions of individuals that contribute to the rich picture of transportation in the DVRPC region. While no model can completely understand the diversity of behavior in a region as large and varied as the greater Philadelphia region, the extensive validation performed on TIM 2 ensures that all significant trends are captured and reproduced. 2010 was used as the validation year.

The TIM 2 model area includes DVRPC's nine member counties plus an extended area of 16 counties in Pennsylvania, New Jersey, Delaware, and Maryland immediately surrounding the DVRPC Region. The number of transportation analysis zones (TAZs) was increased from 2068 (TIM 1.0) to approximately 3550 (TIM 2.0). New zonal boundaries were created to improve the accuracy of the travel model and comply with 2010 census boundaries. The TIM 2 model has four separate time periods – AM peak, midday, PM peak, and night-time, as compared to three time periods modeled in previous models.

The TIM 2 model features a more accurate representation of transportation supply in the DVRPC region. Highway and transit networks were developed from scratch using internet 2.0 open data sources – www.openstreetmap.org for the highway network and Google's General Transit Feed Specification for the transit network. The highway network includes all local streets and integrated with the pedestrian and bike facilities. The transit network is accurate in terms of stop locations, route alignment, service schedules, and fare information, as provided by transit operators.

The demand model was also developed from scratch, with added segmentation by trip purpose and income class. A total of ten trip purposes are modeled to represent internal person travel. Trip distribution uses a gravity model and takes joint highway and transit logsums. The amount of travel between various places, therefore, is dependent not only on ease of travel by car but also the ease of travel by transit. A nested logit model is used for mode choice, where the transit nest is further divided by access mode (i.e., walk and bike access vs. drive access). Park and ride locations for drive to transit trips are determined by a convolution of the highway skim and transit-walk matrices and consider parking lot capacity.

The highway assignment uses VISUM's LUCE algorithm, an origin-based gradient equilibrium method. LUCE not only improves run time and convergence but also provides better path flow results than the path-based method used in TIM 1.0, at the cost of increased memory use. The transit assignment (transit sub-mode and path choice) uses VISUM's timetable based assignment, considering the real-world schedule including the transfer time. Beyond traditional transit forecasting, it is useful for analyzing the effects of operational improvements and time-table changes. The model is currently used for various highway and transit forecasting and analysis projects since 2013.

TIM 3.0

Many new travel demand models use the new Activity Based (AB) paradigm. DVRPC is currently developing an AB model, which will be designated as TIM 3.0.

Land Use Model

UrbanSim

UrbanSim is a land use model that allocates forecasted growth to small area geographies in the region. It simulates individual decision-makers—such as households, employers, and real estate developers— within the regional real estate market, while incorporating interactions between land use, transportation, the economy, and the environment.

DVRPC uses UrbanSim LLC’s UrbanCanvas cloud-based, census-block level platform. As a key input, DVRPC staff creates a regional development pipeline of proposed, planned, and permitted development projects. The platform allows for data sharing with DVRPC’s planning partners for their review and comments on both data inputs and model outputs. DVRPC utilizes UrbanSim for scenario planning, population and employment forecasting, and other regional analyses.

PTV Modeling Platforms

VISUM

VISUM is the current software platform for DVRPC's regional transportation model. As such it is the tool used to generate, maintain, and edit the network model and data elements which are used by the demand model, as well as the primary tool for sequencing the many interdependent steps of the demand model itself. VISUM 12.5 is the software version used by DVRPC for its TIM 2 travel models. TIM 3.0 will use VISUM 14.0.

VISSIM

VISSIM is micro-simulation software used to model at high resolution, down to the scale of individual vehicles and even pedestrians. Micro-simulation is a method for evaluating the localized impacts of proposed improvements such as adding lanes, reconfiguring intersections and ramps, or even the potential impact of new development.

More information about PTV modeling software can be found at: http://www.ptvamerica.com/

Air Quality Conformity

Transportation conformity is the process by which Metropolitan Planning Organizations (MPOs) or Departments of Transportation demonstrate that transportation projects included in a region's Long-Range Plan (Plan) or Transportation Improvement Programs (TIP) do not cause new air quality violations, worsen existing violations, or delay timely attainment of the National Ambient Air Quality Standards (NAAQS). Transportation conformity is a requirement of the Clean Air Act (CAA) in areas that do not meet the NAAQS or have previously been in violation of the NAAQS. Areas currently not meeting the NAAQS are known as nonattainment areas, and areas that previously have attained the NAAQS are known as maintenance areas.

To analyze conformity impacts, DVRPC employs software developed by the Environmental Protection Agency (EPA) which uses outputs from the regional travel demand model as inputs for estimating changes in emissions.

MOVES

EPA's Office of Transportation and Air Quality has developed the MOtor Vehicle Emission Simulator (MOVES). This new emission modeling system estimates emissions for highway sources covering a broad range of pollutants and greenhouse gases. MOVES is the state-of-the-art upgrade to EPA's modeling tools for estimating emissions and replaces the previous model, MOBILE6.2. Federal regulations require DVRPC to use MOVES2010 software for air quality analysis and conformity determinations beginning in 2013 and MOVES2014 for analysis starting in 2016 or later.

Information about MOVES, including policy guidance on when it must be used for State Implementation Plans and transportation conformity determinations, is available on the EPA's website:
http://www.epa.gov/otaq/models.htm

Air Quality Partnership
Annual Report
Connections 2050
Infrastructure Investment and Jobs Act (IIJA)
Economic Development District