While complex modeling solutions can provide valuable insight, the ease of use and practical applications can often be lost in travel demand models. Models are often developed to give the correct answer, but not to serve as useful tools for future studies. Meanwhile, small- and medium-sized Metropolitan Planning Organizations (MPOs) often do not have budgets or in-house expertise to develop or manage travel demand models. This can create a dependency on the original developer to apply the model and interpret results, which decreases efficiency and increases costs for any project involving the tool.

This is why the Streamlined Travel Model Structure (STMS) was developed. With a focus on providing the end-user with a user-friendly tool, the STMS is a three-step trip-based travel demand model developed in TransCAD. The STMS has been designed to have the flexibility to easily transfer the model structure to different regions without extensive familiarity of Geographic Information System Developer's Kit (GISDK) scripting language.

While the STMS offers an industry-standard and customizable model structure that provides outputs by time-period as well as overall performance metrics and quick mapping capabilities, the primary benefits are the reduced time and cost of model development, as well as the ease of use for the end-user. Customizable inputs, such as time-periods or trip purposes, used by the model can be edited outside the script. Data fields are streamlined to prevent confusion. Adjustments to the TAZ structure do not require a painstaking process of adjusting other inputs or editing the script. This flexibility makes the procedure easily transferable and customizable. The model can be run using a simple and user-friendly user-interface.

The STMS has been developed and deployed by HDR staff for multiple small MPO area models and continues to be enhanced over time. While additional customization of the STMS is always an option, the current flexibility of the STMS reduces the time and cost burden of developing a travel demand model and makes the model more accessible to non-modelers while still allowing for accurate outputs. 

Travel demand models are essential tools for understanding and forecasting daily travel patterns within a region. These models account for various travel markets, including household trips, visitor travel, commercial vehicle movements, and through traffic. Traditionally, data for these models are sourced from specialized surveys targeting specific travel markets, such as household travel surveys, establishment surveys, and external surveys. However, intra-regional travel by visitors remains an elusive segment due to the inherent challenges associated with data collection. Conducting dedicated visitor surveys is often infeasible because of obstacles such as the absence of a defined sampling frame, logistical difficulties in recruitment, and the impracticality of large-scale execution. While establishment surveys may capture some visitor data through intercept interviews, this information is typically insufficient to represent the full spectrum of visitor travel accurately.

The advent of location-based services (LBS) data has created new opportunities to assess visitor travel within a region. LBS data, derived from mobile devices, enables reliable identification of visitor status by inferring home locations, thereby distinguishing resident travel from visitor travel. However, it is commonly known that the LBS data suffers from various biases and often requires benchmarking and adjustment prior to its application for modeling purposes. This study explores two approaches to estimate the magnitude of visitor travel within a region by leveraging LBS passive data alongside ground-truth data, such as traffic counts and vehicle-miles traveled (VMT). The results from these approaches are intended to validate against each other to best estimate the visitor travel.

One approach focuses on isolating the VMT attributable to visitor travel. This method involves comparing the regional total VMT extracted from HPMS reporting database with the estimated VMT contributed from all other known travel markets, allowing for the estimation of the residual VMT associated with visitors only. The VMT estimated in this way for visitors is used to benchmark the derived VMT from visitor trips in LBS data, which is then adjusted accordingly to become the final visitor travel estimate. In study areas where household travel survey data is available, comparing estimated VMT derived from the survey with the estimated resident VMT from LBS data offers an additional means to adjust visitor-contributed VMT and visitor trip estimates.

Another complementary approach compares observed traffic counts at external stations with the vehicle trips from LBS data routed through the external stations and estimates the adjustment factor to be applied to all LBS trips, including those identified from visitors. This method assumes that the nature and magnitude of biases from LBS data are consistent across all travel market segments it represents.

The proposed methodologies were applied to several study areas in Texas to support regional modeling applications. Despite the simplicity of the analysis methods, the results and findings from these study cases are promising, reasonably accounting for the specific regional contexts and unique economic characteristics of each study area.

Traditional travel demand model (TDM) provides detailed representation of regional travel demand, but it tends to be resource intensive and involves significantly long development cycle. Thus, traditional TDM may not suit the use cases for low or no growth areas, which could be served by quick-response modeling and scenario analysis. To this effect, the mobility data products from passively collected Big Data have shown promise as an alternate source to support development of such a quick response modeling tool. This abstract describes the development of a Quick Response Model (QRM) prototype by leveraging Big Data complemented with insights from historical travel surveys and other readily available data sources. 

The developed QRM framework starts with a web-based data application that processes and refines the passive Big Data sources using cloud infrastructure (AWS) to produce seed trip tables representing different travel market segments for a region of interest. These market segments represent travel by population groups such as residents/visitors, travel modes such as auto/truck, and their regional movements such as those made within a region, entering/exiting the region and those passing through. The ability to segment the travel into subgroups and benchmarking/validating them either individually or as higher-level groups against separately prepared data sources is critical for assessing the regional impact from these travel segments both separately and collectively. In this way, the QRM mimics much of the analysis sensitivity and capacity from the TDM framework.

The QRM further refines the seed trip tables developed from passive data sources utilizing a sophisticated Origin Destination (O/D) Matrix Estimation procedure that calibrates the O/D flows to traffic counts, zonal marginal controls as well as regional Vehicle Miles Traveled to arrive at solid base year estimates. The calibrated base year O/D flow is then used, in conjunction with travel impedance information, to estimate gravity trip distribution models.

To serve the forecasting and/or scenario analysis needs of QRM, regional socio-demographics and highway networks from both the base year and the scenario case are used as input to the estimated gravity trip distribution model to develop synthesized base year and scenario trip tables.  The difference between these trip tables is applied as the net change of regional travel demand to the base year trip table to become scenario trip table. In this way, QRM enables its forecasting and scenario analysis capacity. QRM uses the same traffic assignment process as in TDM to load the regional vehicle trips onto the highway network and develop network performance metrics. 

The above summarized QRM prototype is demonstrated for Wichita Falls, Texas as a case study area, as it also has a well-calibrated operating TDM.  A comparison of the results for Wichita Falls from the developed QRM and TDM were favorable at regional, zonal and network link levels. This case study demonstrates the feasibility of utilizing passive data based QRM as an option for quick turnaround modeling analysis.

Small and medium sized areas that experience a faster rate of growth and unexpected mobility outcomes are often not served well by regional models. Cities and counties that surround the urbanized core are often included in regional travel demand models but miss the network and demographic granularity needed to capture localized traffic and land use conditions related to the fast pace of development.  As an example, several development plans can be approved at the same time, without consideration of their collective impacts on the road network, mobility, and air emissions like greenhouse gas.  Through passively collected movement data combined with better scaling capabilities of modeling systems, the process to rapidly deploy models to address more localized traffic impacts is streamlined and efficient.  

This paper dives into this streamlined process of creating a travel model in Sherburne County, MN. The model leverages big data to establish baseline traffic characteristics, a customized zone system, and internal/external trip patterns. This practical approach showcases how modeling is rapidly advancing in sophistication and cost effectiveness. The model supports planning activities such as lane closures, congestion management, and land use impacts.  

Discussion will focus on the following topics regarding the model: 

• Development, including big data collection and integration 
• Application - a case study of a transportation management plan (TMP) scenario