Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
127 4 INTRODUCTION BACKGROUND Travel models are analytic tools that provide a consistent framework with which to understand the effects of transportation, land use, and demographic, economic, and policy changes on transportation system performance. Traditional 4-step or trip-based travel models are composed of a series of subcomponent models that indi vidually address aspects of travel demand and supply, such as trip generation, distribution, mode choice, and route choice, and which collectively generate estimates of travel demand choices and transportation system performance. While trip-based models have been applied extensively over the past 40 years, there is an increasing recognition that these models are unable to represent the dynamic interplay between travel behavior and network conditions and, as a result, are unable to reasonably represent the effects of transportation policies such as variable road pricing and travel demand management strategies. This recognition has led to interest in developing integrated dynamic models that link advanced activity-based demand model components with dynamic network traffi c assignment model components. Activity-based models and dynamic network models have evolved in recent decades and offer the opportunity to overcome many of the limitations of traditional trip-based models. Activity-based models consider individual and household travel choices using a consistent framework that includes an explicit representation of timing and sequencing of travel, using tours and trips as fundamental units of travel demand, and incorpo- rating interrelationships among many long-term and short-term dimensions of travel. There are numerous examples of the successful implementation and application of activity-based models in large metropolitan regions (Vovsha et al. 2004). Dynamic network assignment models were created to address defi ciencies in more traditional static network assignment models. Note that the more generic term ânetwork assignment modelâ is used in this guide to include both dynamic network assignment models and dynamic transit assignment models. Because static assignment
128 Part 2: ISSUES IN ADOPTING INTEGRATED DYNAMIC MODELS SYSTEMS models use broad time periods, typically multihour and in some cases even daily time periods, they are unable to capture the impact of travel demand on network perfor- mance for shorter time periods. Static assignment models also do not adequately repre- sent important network operational attributes like capacity and do not represent traf- fic dynamics such as the buildup and dissipation of congestion (Lawe et al. 2011). As a result, they may provide unrealistic estimates of transportation system performance. Integrated models that incorporate both activity-based and dynamic network assignment components can capture the interplay between travel behavior and net- work conditions and provide greater policy and investment analytic capabilities. By incorporating greater temporal and spatial detail, as well as by better reflecting the heterogeneity of users of the transportation system, integrated dynamic models can better represent the effects of pricing alternatives, transportation systems management and travel demand management strategies, and capacity improvements; offer more robust aggregate forecasts; and provide more detailed outputs to inform investment, air quality, and equity analyses. But despite the recent advances in activity-based demand models and dynamic network assignment models and their increased adoption by transportation planning agencies, there are very few examples of integrated dynamic models that include activity- based and dynamic network assignments components. The limited number of examples is likely a result of integrated dynamic network model costs and development schedule, data requirements, institutional issues, and software and hardware requirements. As a result, the potential benefits from these recent advances have not been fully realized. Two notable successes in integrating these advanced models are the recent SHRP 2 C10 projects; Projects C10A and C10B projects have established integrated activity-based and dynamic network assignment model systems and subjected these model systems to a set of validation and sensitivity tests. In addition, a number of MPOs have recently embarked on integrated dynamic model development efforts, although these efforts are still under way. These examples represent the first forays into a rapidly maturing field toward which industry practice is moving. These model systems demonstrate the capabilities of the new integrated dynamic network model paradigm and also provide instructive information about the challenges faced when developing and applying these new model systems. PURPOSE The purpose of Part 2 is to examine the practical issues that MPOs, state DOTs, and other transportation agencies face if they are considering migrating from traditional to advanced travel demand forecasting approaches using SHRP 2 travel demand forecast- ing products. In this part of the guide, advanced models refer to model systems that incorporate activity-based travel demand models and dynamic network models, and in which these primary components exchange information in a systematic way to reach a stable solution. These integrated dynamic models are of interest because they can provide a common analytic framework with which to evaluate a wide range of plan- ning and operational strategies to address local and regional goals (Resource Systems Group et al. 2014).
129 Chapter 4: INTRODUCTION Advanced models as described represent an area of travel behavior research and practice in which theory, methods, and tools are rapidly evolving. Because of the dynamic nature of this field, this part of the guide identifies general implementation challenges and potential next steps for addressing these challenges and is intended to inform efforts by state DOT, MPO, and other transportation agency staff as they con- sider and pursue the development of integrated dynamic models. In order to understand the practical issues associated with integrated dynamic models, the authors begin with four case examples that briefly summarize integrated dynamic models that have been developed or are under development. While there are numerous academic research efforts as well as other regional integrated dynamic model development efforts currently under way, the four case examples described are distinct because they have all been used or will be imminently used to evaluate investment and policy alternatives. However, not all model systems are at the same level of development. The integrated model systems developed as part of the SHRP 2 C10A and C10B projects are fully integrated in that both model systems incorpo- rate activity-based models systems and regional-scale dynamic network models that exchange information in a systematic way. The model systems developed for the San Francisco County Transportation Authority (SFCTA) and the Maricopa Association of Governments (MAG) are only partially integrated. The SFCTA model system includes an activity-based model and an all-streets network, but this network covers only a portion of the entire region for a portion of the day, and network performance mea- sures from the dynamic network model are not fed back to the activity-based demand model. Similarly, although the MAG dynamic network model is very large, it does not cover the entire region or entire day, and the network performance measures from the dynamic network model are not fed back to the activity-based travel model. Following the case examples, there is an examination of critical integrated dynamic model development considerations, including development costs and schedule, data requirements, application challenges, institutional issues, and software and hardware requirements.
