Guide for Identifying, Classifying, Evaluating, and Mitigating Truck Freight Bottlenecks (2017)

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.

9 C h a p t e r 1 This Guidebook provides state-of-the-practice information to transportation professionals on identifying, classifying, evaluating, and mitigating truck bottlenecks. The bottleneck analysis described in this Guidebook is focused on utilizing truck probe data rather than traditional travel demand models. The primary application for the methodologies is evaluation of truck bottlenecks for prioritizing investment decisions. This Guidebook serves the following purposes: • Defines a common language related to truck freight bottlenecks; • Classifies truck freight bottleneck categories based on causal and contributing factors; • Describes truck bottleneck state-of-the-practice; • Provides highlights from several case studies related to truck bottlenecks; • Describes data sources used for truck bottleneck analysis; • Provides a spatially scalable methodology for identifying truck freight bottlenecks; • Describes quantitative measures for truck freight bottleneck categories for determining bottle- neck severity, impact, and ranking and subsequent decision making; • Describes mitigation options for truck freight bottlenecks; and • Describes how to integrate truck freight bottleneck analysis into the planning process. Examples of truck bottleneck analysis and notable practice highlights are provided throughout the Guidebook. The Guidebook is intended for two primary audiences: 1. Transportation planners that are conducting freight-related analysis or developing freight- related planning documents and 2. Research and operational staff that are interested in developing freight bottleneck analyses relevant for transportation planning processes. 1.1 Key Themes in Truck Bottleneck Analysis There are a number of overarching themes and observations related to the state-of-the-practice in truck bottleneck analysis. Highlights of these observations are listed in this chapter to give practitioners an overview of key issues related to truck bottlenecks. 1.1.1 Classification Structure Is Needed Truck bottleneck classification is not an exact science. There is a need for the development and clarification of a truck bottleneck classification scheme. Many of the resource write-ups in Appendix B are associated with “classifying bottlenecks.” While on the surface it appears there are many examples available, upon review of the references, there are often bottleneck terms Introduction

10 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks used interchangeably or other nomenclature issues that could be remedied with a uniform clas- sification structure (as introduced in Chapter 2). 1.1.2 Identification of Truck Bottleneck Cause Not only is truck classification a challenge, there is not always clear identification of truck bottleneck cause. Many studies identify or evaluate truck bottlenecks and rank specific locations (typically with some form of a delay measure), and then a secondary (project-level) analysis and/or other data sources are necessary to identify key issues/problems that may cause a truck- specific bottleneck. In practice, there are typically project-level quantitative and qualitative evaluations needed to identify truck bottleneck causes. This secondary project-level analysis is an element of the truck bottleneck analysis process described in this Guidebook. 1.1.3 Connecting Mitigation Strategies for Specific Truck Bottleneck Causes There often is not a clear quantifiable link between mitigation strategies and a specific bottle- neck cause. For congestion mitigation, this may not be a concern as mitigation strategies that alleviate congestion for all vehicles also benefit truckers. However, there is a need to quantify the benefit of bottleneck improvements to truckers, particularly for situations due to restric- tions (i.e., geometric or height restrictions or truck bans). The Guidebook proposes a method for doing this in Chapter 6. 1.1.4 Truck Bottleneck Analytics Are Generally Consistent and Scalable There are a number of practices in the literature related to facility-based mobility analysis that include a truck component (e.g., ranking roadway sections by truck delay per mile). These practices generally integrate speed and volume data sources, and these practices are scalable from roadway sections to longer sections to urban area or statewide analyses. 1.1.5 Trip-Based Versus Facility-Based Analysis Many of the travel speed and congestion-related bottlenecks analyze particular segments or facilities. Congestion measures such as delay, travel time index (TTI), or planning-time index (PTI) (reliability) are then ranked for the corridors. However, the trucking industry is more concerned about trips and delivering goods from point A to point B. In some ways, a facility-based analysis approach misses the trucking deci- sions that are part of the origin-destination decisions that truckers must make. There is a need for analytics that consider the origin-destination pairs and evaluates trip planning and specific routes in comparison to one another. Methods for doing these analyses are described in this Guidebook in Chapter 5. The authors of this Guidebook believe that understanding how to manipulate the increasingly ubiquitous probe data sources for trip-based analysis will become more important in the future as these datasets become even more prevalent and computing power and computing knowledge increase. This dynamic can be illustrated through considering a speed analy- sis for a corridor. An analysis could focus solely on the speeds on the Case Study Highlight Over the past several years, Transport Canada has developed a freight fluidity measure. The measure is multimodal. Transport Canada has developed an integrated supply chain tool that measures individual segments of the supply chains as well as end-to-end transit time of freight flows. Over time, Transport Canada has obtained supply chain data from multiple modes, including ocean, as well as port-related, rail, trucking, air and logistics and warehousing to power the fluidity measure. More details are provided in Appendix B.

Introduction 11 corridor or an analysis can be considered for a corridor and its parallel facilities. In this second case, the analysis is much more similar to an analysis that examines the travel time between the initial and termination points of a corridor. The determination of the bounds of the analysis is often influenced by the perspective of the party conducting the analysis. A state DOT may only look at state-owned roads without consideration of local roads. A metropolitan planning orga- nization (MPO) may only examine roads within its jurisdiction rather than alternative routes that may be outside of its jurisdiction. 1.1.6 Truck-Specific Data Sources The transportation industry has benefitted greatly in recent years from the increasing abun- dance of probe speed data. However, the user must clearly understand this data source and what implications it can have on a truck bottleneck analysis. Ideally, truck-specific speed data would be obtained and, depending upon the truck bottleneck application, speed data specific to single- unit and/or combination-unit trucks may be desirable. The breadth of coverage of the speed data also needs to be considered. For example, coverage of first-mile and last-mile connectors is typi- cally important in speed analyses of regional networks, so there needs to be special examination of the speed data set to confirm that these roadways are included. Similarly, truck volumes are needed to combine with the speed data to create truck delay statistics. Truck volume sources can be local automatic traffic recorders, weigh-in-motion sites, planning models, and/or even the Federal Highway Administration (FHWA) Highway Perfor- mance Monitoring System (HPMS). The key is using the best available truck-specific data for the truck bottleneck study. A specific area of concern is the use of average daily truck volume values. These may be appropriate for some generalized analyses. However, when analyzing nonrecur- ring truck delay, a more discrete truck volume set is needed. For example, to examine the delay impact of a crash, it is ideal to obtain the specific truck volume data that occurred at the time of the crash. Alternatively, the needed truck volume data can be estimated using the annual average daily traffic (AADT) value combined with seasonal, daily, and hourly factors related to the type of roadway where the crash occurred. Information on truck factors can be found in the Highway Capacity Manual (1). These nuances and data sources are described in more detail in Chapters 4 and 5. 1.1.7 Computation of Reliability Measures There are a number of possible sources for “all vehicles” speeds or even truck-specific speeds. The industry would benefit from recom- mendations on what reliability measures are most useful for truck bottle neck analyses, computational procedures, and weighting by truck vehicle miles traveled (VMT), including temporal and spatial aggre- gation guidance. Details for computing truck reliability measures are provided in Chapters 5 and 6, along with Appendix D. 1.1.8 Engaging Trucking Stakeholders Many of the resources related to truck bottlenecks relied upon engag- ing truck companies and associated stakeholders. These stakeholders are intimately familiar with the roadway shipping lines and impediments that impact their daily schedules—they are a key resource for public agency professionals. Practitioners will ideally remember to engage this Case Study Highlight Recent work by the Virginia Department of Transportation (VDOT) identified truck bottlenecks throughout the state. A novel approach in the study was the interviewing of over 180 stakeholders representing manufacturing, distribution firms (truck firms, wholesalers, etc.) and an assortment of retail, mining, agricultural and other firms. Respondents indicated congestion was the most prominent concern, followed by the driver shortage and then high fuel costs. The predominant solution proposed by respondents was some form of added capacity. More details are provided in Appendix B.

12 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks valuable stakeholder when identifying truck bottleneck locations as well as mitigation strategies. Engaging truck stakeholders is particularly important for first-mile and last-mile connectors, where truck speed data are typically less available. This engagement can occur through conven- ing large group meetings, one-on-one interviews, or electronic survey processes. TRB’s NCFRP Report 25: Freight Data Sharing Guidebook (125) has guidance on obtaining information and data from freight organizations. 1.1.9 Mapping Tools Are Effective at Illustrating Truck Freight Bottlenecks Several resources were found that included interactive maps and/or analytics to provide infor- mation for truck bottlenecks investment decisions. These mapping and GIS tools really help to tell the truck bottleneck story to decision makers and policy makers. 1.2 Classifying Truck Bottlenecks This Guidebook embraces a broad term for “truck freight bottlenecks” as any condition that acts as an impediment to efficient truck travel, lead- ing to travel times in excess of what would normally occur. This defini- tion encompasses a wide range of events and conditions, all of which add time to the delivery of truck freight shipments, from the time those shipments leave their origin to the time they arrive at their destination. This broad view starts with the general understanding of the term “bottleneck”—a place where traffic congestion routinely forms. This rou- tine congestion may be caused by a lack of roadway capacity for the typical peak traffic volumes on that road section (commonly called “recurring congestion” in the literature). The definition of truck delay is extended in this Guidebook to include factors other than traffic congestion that increase the travel time for truck trips. These additional factors include issues such as: • Additional trip distances caused by deficient bridge design (height, weight, width, etc.); • Additional miles caused by load restrictions, whether seasonal weight limits or for hazardous materials; and • Truck processing delays at sites such as weigh stations, border crossings, marine terminals, rail yards, warehouse/distribution centers, etc. Classifying truck bottlenecks needs to occur first due to different analysis methods for travel speed-based and process-based truck bottlenecks. The suggested bottleneck classification is designed to describe locations that add travel time to truck trips while simultaneously describ- ing the causes of those delays because the causes of the delays relate directly to the options for eliminating or mitigating them, and thus eliminating or mitigating the delay itself. The following outlines bottleneck classifications: • Travel speed- and process-based: Are the bottlenecks caused by congestion and travel speed limitations or increased VMT? • Recurrent and nonrecurrent: Is the bottleneck a daily occurrence? Table 1-1 and Figure 1-1 provide key characteristics of travel speed-based delay. Travel speed- based delay is defined as locations where delay occurs as a result of oversaturated traffic condi- tions, temporary loss of operation capacity, or because roadway design causes truck-only delays. Case Study Highlight The University of Maryland Center for Advanced Transportation Technology (CATT) Laboratory Vehicle Probe Project Suite is an example of a suite of visual tools and dashboards to support operations, planning, analysis, research, and performance measures using probe data in concert with other agency transportation data. http://www.cattlab. umd.edu/?portfolio=vehicle-probe- project-suite.

Introduction 13 Source: FHWA Office of Operations, Traffic Congestion and Reliability: Trends and Advanced Strategies for Congestion Mitigation, September 2005. (126) Non-Recurring (e.g., Special Events) 5% Bottlenecks 40% Traffic Incidents 25% Work Zones 10% Weather (Snow, Ice, Fog) 15% Poor Signal Timing 5% Recurring Congestion Figure 1-1. Sources of delay for all vehicle types (trucks and autos), national level, all vehicle types. Cause of Travel Speed-Based Bottleneck Bottleneck Type Truck bottlenecks caused by simply too much traffic volume Peak-period traffic Roadway geometrics (lane drop) Steep grades/terrain Special event traffic Seasonal traffic volumes Surge truck traffic from unloading of large container ships Truck bottlenecks caused by temporary loss of operational capacity Work zones Weather Poor signal timing Traffic incidents Processing delays (toll booths, weight enforcement stations, terminal gates, international border crossings) Truck-only bottlenecks (delays) caused by roadway limitations due to vehicle characteristics Roadway geometrics Steep grades Tight curves Narrow lanes Table 1-1. Classification of travel speed-based delay truck bottlenecks.

14 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks Impact of Process-Based Bottleneck Bottleneck Type Rerouting Low bridge heights Truck weight restrictions Hazardous materials restrictions Making additional trips Spring thaw load restrictions when no alternate routes Truck size (length) and weight restrictions Truck bans or restrictions Time-of-day restrictions Truck pick-ups and deliveries in off-hours Truckers having to search/wait for loading zones/parking Having to make inefficient movements such as circling a block, because the last-mile facilities (e.g., parking, load zones, terminal gates) are not suitable, lack capacity or poorly managed Table 1-2. Classification of process-based delay truck bottlenecks. These delays can also be caused from processing activities that occur at key freight locations. Table 1-2 provides key characteristics of process-based delays. Process-based delay is defined as locations that force trucks to use lon- ger, more circuitous paths than passenger cars would take if making the same trip, delays at specific locations related to freight, such as terminal gates, or requirements that trucks carry less cargo than they would otherwise carry if not legally restricted. The suggested bottleneck classifications start with “travel speed (typically roadway congestion)” bottlenecks (recurring), because those delays are shared with cars and, therefore, the benefits from improve- ments made to mitigate those delays will be viewed differently by agen- cies funding the required mitigation. The travel speed-based bottlenecks are further divided into three subcategories: 1. The first subcategory is locations where congestion forms primarily as a result of too much base traffic volume. 2. The second subcategory is locations where “temporary” operational limitations decrease operational capacity below traffic volume levels that would otherwise be able to operate without congestion. 3. The third subcategory of travel-speed bottlenecks is where only trucks are slow because of their larger size and performance characteristics reduce their mobility on a road as compared to cars. These bottle- necks are due to roadway geometrics (grades, tight turns, narrow roads) that are difficult for trucks. Notable resources related to classifying bottlenecks are described in Appendix B and include: • An Initial Assessment of Freight Bottlenecks on Highways. • Quantifying the Contributing Factors of Traffic Congestion Using Urban Congestion Report Data. • Oregon State Highway Performance Data and Metrics Related to Freight. • Positioning Hampton Roads for Freight Infrastructure Funding MAP-21 and Beyond. • Freight bottlenecks in the Upper Midwest: Identification, Collaboration, and Alleviation/ Identifying and Characterizing Truck Bottlenecks in the U.S. Mississippi Valley Region. • Oregon Department of Transportation (ODOT) Region 1 Corridor Bottleneck Operations Study.

Introduction 15 The second broad category of truck bottlenecks encompasses operational process-related delay situations in which the attributes of the trucks, or the cargo they carry, result in travel times longer than passenger vehicles traveling from the same origin to the same destination would experience. Low bridge heights, truck size/weight restrictions, terminal queues, and truck bans are a sampling of examples that cause operational process-related delays. Definitions were also developed for “Identifying,” “Classifying,” “Evaluating,” and “Mitigat- ing” bottlenecks to guide the proper identification and categorization of the selected case studies, many which are discussed in this Guidebook. The following definitions were used: • Identifying Bottlenecks. Locating where bottlenecks are in the transportation system based on qualitative and/or quantitative methods. • Classifying Bottlenecks. Associating a cause to the truck bottleneck. • Evaluating Bottlenecks. Estimating the extent, duration, and/or severity of the truck bottle- neck; sometimes this is augmented with bottleneck rankings and can be part of the identifica- tion process or a separate (more detailed) analysis. • Mitigating Bottlenecks. Exploring potential truck bottleneck(s) solutions or analyzing existing mitigating efforts. 1.3 Overview of Truck Bottleneck Data Considerations Data that can aid in determining the causes of bottlenecks are as follows: 1. Collision Data. How regularly do incidents occur on a specific corridor? For reliability pur- poses, have trucks rerouted to avoid uncertainty created by high-incident locations, and if so, what type of additional time and/or VMT are associated with the alternative route? 2. Weather Data. Are there seasonal travel pattern differences that create bottlenecks? (Consider high-incident locations). 3. Freight Facility Gate Data (ports, rail yards, intermodal facilities, border crossings, at-grade railroad crossings, etc.). Is the data capturing the peak months for goods movement? 4. Special Event Data. How does special event traffic impact truck corridors? 5. Work Zones Data (closures, detours, reduced lane widths, rough pavement, speed restrictions, etc.). Are there higher incident rates or longer travel times due to detours? 6. Operational Restrictions Data (time-of-day delivery restrictions, peak-hour fees, etc.). Are there hours of operations restrictions related to noise ordinances, limited gate hours at ports, curbside parking restrictions, higher toll rates, and/or other impediments that add to the travel time? 7. Truck Parking Data. The availability and usage of truck parking can provide information, particularly in metropolitan areas, on the origin-destination pairs for trucks. Additionally, the lack of sufficient truck parking causes trucks to add mileage to their trips as they search for parking. This is a form of process-based delay. 8. Roadway Features Data (grades, lane widths, turning radii, signage, pavement/striping/ markings condition, etc.). 9. Data based on input from the trucking industry and transportation agency staff. The ultimate use/application of the output from an analysis drives the data processing pro- cedures, as well as data collection and data reduction decisions. The primary application for the methodology discussed in this Guidebook is the determination of truck bottlenecks for prioritiz- ing investment decisions. While that sounds relatively straightforward, there are still important considerations for the data analyst that will impact data collection, data reduction, and data processing steps.

16 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks It is not always possible to obtain data at the spatial and temporal granularity for the specific location(s) of interest. Table 1-3 illustrates the spatial and temporal data availability tradeoffs that are rather commonplace in performing truck bottleneck studies when complete data are not available. Note that these tradeoffs are applicable for both volume and speed data. Speed (or travel time) and volume data are the most common (and critical) for truck bottleneck analyses. 1.3.1 Guiding Principles of Truck Bottleneck Analysis A variety of technologies and methods are used to estimate truck bottlenecks. The methods generally include direct measurement of travel time and delay, derivation (virtual probes) of travel time and delay, and a combination of direct measurement and model use. While more detailed analysis procedures are provided in Chapters 5 and 6 and Appendix D for all the topical areas touched upon below, this section is meant to simply provide some guiding principles and general overview (2) to familiarize the reader with key concepts before more details are provided in later chapters. In light of the literature and current practice, the following criteria are established that a mea- surement procedure should meet: • Congestion performance should be primarily assessed from the user’s perspective, not the facil- ity’s. Travelers experience the whole trip; isolated portions of it influence trip performance but the whole experience is important to travelers. This criterion implies that travel times be the basis for performance measures for congestion. Using travel times also is consistent with how freeway performance measurement is conducted and travel times resonate with the general public; they are easy to communicate. • The best way to develop travel times is to measure them directly. Technologies that track individual vehicles accomplish this, as do agency probes. Global Positioning Systems- (GPS-) based methods may or may not; these currently are used by private vendors who employ pro- prietary data reduction methods, and it is difficult to know if they develop travel times from Data Availability Spatially Temporally Most desirable are… ….actual data for the specific site(s) of interest… …and/or data at desired time granularity to satisfy the application (e.g., annual, hourly, 15-minute, 1- minute). Less desirable are… …estimated data from similar site(s)… …and/or data aggregated over time because desired granularity not available. Source: Margiotta, R., B. Eisele, and J. Short. Freight Performance Measure Approaches for Bottlenecks, Arterials, and Linking Volumes to Congestion Report, Federal Highway Administration, Report No. FHWA-HOP-15-033, Washington, D.C., August 2015. Available: http://www.ops.fhwa.dot.gov/publications/fhwahop15033/fhwahop15033.pdf. (2) Table 1-3. Speed and volume spatial and temporal data availability considerations.

Introduction 17 tracking individual vehicles over a distance or use instantaneous vehicle speed measurements. If vendors ever develop data based on true origin-destination traces for individual vehicles, then directly measured travel times will be available. • Travel times should be measured continuously—or nearly so—to develop distributions of travel times. Having access to the complete travel time distribution allows the calculation of reliability and provides a more complete picture of performance. • Delay at individual signals, or at other specific bottleneck locations along a corridor, should be measured. The ability to identify specific bottlenecks along a corridor is a vital step in per- formance management. Therefore, a “drill-down” capability to identify where problems exist, once the performance of the arterial corridor is established, is needed. 1.3.2 Corridor-Wide Travel Time Data Reduction After a distribution of travel times is established, a wide variety of performance measures can be created. The first step in developing corridor-wide measures is to work out the segmentation of the corridor so that the data can be properly reduced. Because of issues of “time-distance displacement” in combining data, the corridor should not be excessively long: 10 miles is a reasonable maximum. (If travel times from multiple segments are added to get the route travel time for a given time period, this will not correspond to the travel time measured from a vehicle’s perspective, which will pass over downstream segments at different times.) Above that, care must be used in interpreting the results. In all likelihood, the corridor (i.e., longer analysis reporting segment) of interest will be longer than the data collection segments that comprise it. Therefore, a method for combining the mea- surements for the data collection segments (e.g., where a reidentification detector is located or the segments on which GPS-based travel times are reported) into the corridor is needed. Four methods can be used: 1. The most direct method is simply to track the travel times of individual vehicles throughout the length of the entire corridor and develop the travel time distribution from them. This currently is only possible with the reidentification technologies. It is the “purest” of the meth- ods as the corridor travel time is directly measured. However, there are problems with this approach: a. Sample sizes may be small, because of vehicles entering and leaving the corridor at differ- ent points. b. Due to the possibility of travelers making intermediate stops at activities along the corridor, some recorded travel times will be excessively long. Statistical procedures have been devel- oped to weed out these long trips, but they are post hoc in nature and may result in exclud- ing sound data. (3) These problems can be minimized by keeping the corridors reasonably short in length, even shorter than the 10 miles recommended above. 2. Develop travel time distributions for each data collection segment first, and then combine to get the corridor distribution. The moments of the distributions for the individual data collection segments are calculated. These include the following metrics for both travel time and space mean speed: minimum and maximum values; 1st, 5th, 10th, 15th, 20th, 25th, 30th, 40th, 50th, 60th, 70th, 75th, 85th, 90th, 95th, and 99th percentiles; mean; and variance. Corridor metrics are simply the sum of the data collection segment metrics. Past research has found that travel times on adjacent segments are not statistically independent (i.e., they are assumed to be correlated), and hence variances and percentiles cannot be added (but means can) (4, 5, 6). Recent work by Isukapati et al. suggests that in practice, they can be additive (7). However, their work is based on examining a single freeway corridor with rela- tively uncongested conditions—the applicability to congested and/or arterial conditions is unknown.

18 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks 3. Develop corridor-wide travel times first, and then create the corridor distribution from them. In this approach, a corridor travel time for each time epoch (e.g., every 5 minutes) is created. These travel times are then the observations in the travel time distribution from which con- gestion and reliability metrics are created. This method avoids any thorny statistical problems with combining distributions and most closely resembles data collected from direct observa- tion of travel times from end to end. 4. Apply the virtual probe or trajectory method. This is not a distinct method but an extension to method No. 3 above, which has the problem of not precisely replicating the passage of vehicles over the facility in time and space. [Method No. 2 also suffers from this time-distance displacement but there is no easy way to address it for percentiles; mean values could be used, however (8).] This is less of a problem for relatively short facilities, such as the recommended 10 miles. However, as trip lengths extend, the problem becomes exacerbated. Based on this assessment of travel time data reduction, the following recommendations are made (9). a. Using the principle that the best way to develop travel times is to directly measure them, Method No. 1 should be the preferred method, but it has limitations because of small sample sizes and interrupted trips. It also is applicable only to the reidentification data collection technologies. Therefore, the preferred approach is Method No. 4, especially for long corridors. Method No. 3 will suffice for corridors that are not longer than 10 miles. b. Adding segment distributions to obtain percentiles, which are the basis for most reliability metrics, is not recommended for facility performance. Serious theoretical questions exist that have not been adequately addressed with empirical evidence, and there appears no simple way of accounting for the time-distance displacement problem with this method. Additional research may override this recommendation or develop adjustments for its application. c. If only mean travel times are desired, then adding mean segment travel times to obtain facil- ity travel time is acceptable. There are several sources of truck origin-destination data that can be used to combine with corridor-specific data. Truck origin-destination data is available through travel demand models. It can also be extracted from commodity flow databases. There are also techniques to develop truck origin-destination data through truck GPS data by tracking individual trucks between sequential locations where the data show them to be stopped for extended periods of time. Figure 1-2 shows a map of truck trip origins and destinations identified in the Atlanta metro- politan region using truck GPS data. Understanding these origin-destination patterns is particu- larly important for situations when there is a desire to consider through-trucks trips relative to internal truck trips or situations where the impacted jurisdictions need to be identified. Under- standing origin-destination patterns is also important for process-based delays to determine the specific type of rerouting that occurs due to restrictions such as a low clearance bridge, weight- restricted roads, or other truck bans. Additionally, with the availability of truck origin-destination data through commodity flow databases and more recently with transactional data, there is the ability to match the corridor- level delay analysis with key elements of larger goods movement patterns. This allows for the impact of truck delays to be better understood within the context of supply chains and broader economic activity.

Introduction 19 Source: Georgia Department of Transportation (GDOT) Freight & Logistics Plan, 2011. (54) Figure 1-2. Truck trip origins-destinations in Atlanta Region identified using truck GPS data.