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.

22 The first step in the truck bottleneck methodology is to identify, collect, quality check, orga- nize, and link the various data sources available to the agency that are needed to identify and quantify bottleneck locations. The more and better the data available, the better the results of the analysis. However, useful results can be obtained with even modest data resources. 3.1 Truck Bottleneck Data Considerations Speed data are available from a number of sources. Based on the state-of-the-practice findings related to the growth of probe-vehicle speed data sources and their use, this Guidebook focuses on how to use probe data to identify truck freight bottlenecks. Availability of probe-vehicle data sources will become more temporally and spatially prevalent in the future. Although probe data are the focus of this Guidebook, it is important that Guidebook users be aware of selected char- acteristics of other available travel time data sources. Table 3-1 provides a synopsis of the major types of speed data collection methods/systems for travel time and selects derivative products. All types except for GPS-based data require that agencies deploy and maintain field equipment. What is notable in Table 3-1 is that probe-vehicle sources are scalable agencywide; they offer the ability to perform truck bottleneck analyses at the roadway, region, metro, state, or even national level. In comparison to other technologies, this scalability is where probe-vehicle sources really shine. This will only improve as these data increase in availability. The most significant issue in terms of scalability for GPS-based data occurs on higher- classification roadways. On these roads, there are often sample sizes that are too small to develop summary information on vehicle speeds. As indicated in Table 3-1, in comparison to other technologies, probe-vehicle methods have smaller sample sizes, which impacts the ability to characterize the travel time distribution. Sam- ple size and travel time distribution is better from sensors in the field because they typically col- lect more detailed samples. As GPS-based data methods improve, the concerns of limited sample size may be mitigated, particularly on higher classification roadways. It is important to note the “virtual probe” travel time option that is discussed in Table 3-1. In this case, the analyst “traces” a modeled vehicle through time and space along a facility of interest to obtain an estimate of travel time through the corridor. From these estimated travel times, the travel time distribution for an entire corridor can be estimated. While this method is good for spe- cific corridors, it can become cumbersome and more complex to apply over large spatial networks. The data types identified in Table 3-1 represent those that are available as of the writing of this report. It is possible that vendors currently offering vehicle probe data based on roadway C h a p t e r 3 Identify and Assemble Data

Identify and assemble Data 23 segments will provide individual vehicle data that allow constructing travel times between ori- gins and destinations (O/Ds). These O/D travel times would be directly measured rather than synthesized. In terms of GPS-based data, evaluations from the University of Maryland and Virginia Center for Transportation Innovation and Research (VCTIR) suggest that the accuracy of these data are questionable on arterial streets that have very congested, oversaturated conditions (multiple cycle failures). Accuracy problems also exist on lower-order functional classes, where probe samples are likely to be small. (10) For the purposes of performance monitoring and bottleneck identification, where the primary interest is in the relative rankings and trend analysis of truck bottlenecks, the accuracy problem is not as severe as for other uses such as traveler information. As vendors gain more experience in collecting and processing travel time data, the accuracy problem may be minimized, but there is no guarantee of that happening. For the moment, users need to be aware of the accuracy problems especially when making benefit estimates. The impact of truck bottlenecks in monetary terms can be estimated by translating bottleneck delay data into dollars. There are several sources of estimates of the cost of truck travel delay such as the FHWA Highway Economic Requirement System and the American Transportation Research Institute’s An Analysis of the Operational Costs of Trucking: 2015 Update (128). The monetary impact of delay can also be estimated based on the type of cargo that is being delayed. Commodity flow data and transactional data can be used for these types of estimates. Technology Sample Size Characterized Distribution of Travel Times Ability to Scale Agencywide Reidentification of vehicles (ALPR, pavement sensors, toll-tag readers) Excellent Excellent Poor Reidentification with MAC address matching (Bluetooth) Good Good Fair GPS-based data (commercial vehicle probe) Fair Fair (but improving) Excellent Virtual probe Excellent Excellent (but derived) Excellent Agency-driven probe vehicles Poor Poor Poor Sources: Remias, Stephen M., Alexander M. Hainen, Christopher M. Day, Thomas M. Brennan, Jr., Howell Li, Erick Rivera-Hernandez, James R. Sturdevant, Stanley E. Young, and Darcy M. Bullock, Performance Characterization of Arterial Traffic Flow with Probe Vehicle Data, Transportation Research Record: Journal of the Transportation Research Board, No. 2380, Transportation Research Board of the National Academies, Washington, D.C., 2013 (127), with “virtual probe” assessment added by Cambridge Systematics, Inc.; and 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). ALPR = automatic license plate readers. Table 3-1. Comparison of travel time data collection technologies, derivative products, and selected data characteristics.

24 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks 3.2 Potential Data Sources At a minimum, data are needed on the performance and use of the road system. That is, how fast are vehicles (and in particular, trucks) moving, where are they being delayed, and how many of them are using the roads and/or being delayed? Once the ability to compute delays can be accomplished, more effective bottleneck analysis can be performed if data also are available that describe the potential causes of delay. These include “temporary” events, such as: • Vehicle crashes and other incidents, • Construction activities, • Bad weather, and • Special events. Not all state agencies and MPOs will have these data available. In addition, for many of the cases where states have these data, they will not be uniformly available for all geographic regions in the state. This is an acceptable circumstance as travel speed bottlenecks can still be identified without these data, but the agencies will find that more work is required to understand the causes of those bottlenecks if these data are not available. Finally, data that describe physical limitations in the roadway infrastructure that can cause delay also are desired. These items include geometric and terrain features that can slow vehicles— especially loaded trucks. Examples of data items [often found in part in state DOTs and MPOs geographical data (GeoData) catalogs] that could be gathered and included in the data system are as follows: • Roadway geometric limitations (e.g., narrow lane widths, low-height bridges); • Grades steep enough to affect truck speeds; • Activities that delay vehicles (e.g., toll booths, weigh stations, international border crossings); and • A lack of truck-specific, last-mile facilities such as parking or load zones. A list of potential data sources for each bottleneck classification type is provided in Table 3-2. By obtaining data on these activities and roadway features and placing them within the truck bottleneck data analysis structure, it is possible to develop automated procedures that allow agencies to not only readily compute the presence, size, and frequency of congestion bottlenecks, but also to obtain good insight into the causes of those bottlenecks. It should be noted that the vast majority of these data are from public sources. While there is much data that exist in the private-sector freight community, the challenges in obtaining, analyzing, and aggregating sufficient data across enough companies typically makes the private sector an inefficient source for conducting a comprehensive analysis. Data from freight trans- actions is becoming increasingly available and can provide detailed information on O/D pat- terns, but they do not provide the temporal or roadway detail that is most useful for bottleneck analysis. Additional information on the use of private-sector freight data can be obtained from NCFRP Report 25: Freight Data Sharing Guidebook (125). 3.3 Description of Key Data Sources 3.3.1 Vehicle Speed and Travel Time Data States and MPOs currently have access to data sets that can provide estimates of where conges- tion is occurring on at least a portion of their roadway system. At a minimum, every state DOT and MPO has access to the NPMRDS made available by FHWA. These data provide estimates

Identify and assemble Data 25 Bottleneck Category Bottleneck Type Example Data Sources Travel Speed- Based Bottlenecks Peak-period traffic State DOT Traffic Count Data Roadway geometrics (e.g., lane drop) and attributes (e.g., tunnels) State DOT Roadway Inventory Database Steep grades/terrain State DOT Roadway Inventory Database Special event traffic State or Regional Visitors and Convention Bureau Seasonal traffic volumes State DOT Traffic Count Data Work zones State DOT Construction and Maintenance Logs Weather National Weather Service Poor signal timing Local Traffic Management Center Vehicle crashes or other traffic incidents State DOT Crash Database Tight curves State DOT Roadway Inventory Database Surge traffic from unloading container ships Port Vessel Schedule Data, Port Activity Data (e.g., PierPass in Southern California) Narrow lanes State DOT Roadway Inventory Database Process- Based Bottlenecks Low bridge heights State DOT Roadway Inventory Database Truck weight restrictions State DOT Roadway Inventory Database Hazardous materials restrictions State DOT Roadway Inventory Database Load restrictions when no alternate routes (e.g., spring thaw) State DOT Oversize/Overweight (OS/OW) Permit Office Truck size (length) restrictions State DOT OS/OW Permit Office Time-of-day restrictions Local Municipalities, Truck Operators Truck pick-ups and deliveries in off-hours Local Municipalities, Truck Operators Node-based delays (toll booths, weight enforcement stations, border crossings) State Highway Patrol, Facility Operators, Local Customs Office Having to make inefficient movements such as circling a block due to unsuitable trip end facilities (e.g., parking, load zones) Local Data Collection Efforts Table 3-2. Potential data sources for each bottleneck classification type.

26 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks of travel times at which vehicles operate on the entire National Highway System (NHS). Other probe-vehicle or sensor datasets can also be used to estimate speed or travel times. The NPMRDS provides estimates of travel times for passenger cars, trucks, and all vehicles combined for each directional segment of the NHS for every 5 minutes of the year. The excep- tion is when no instrumented vehicles report using those segments during that five-minute interval. In that case, the NPMRDS provides no estimate of the road’s performance at that loca- tion for that time interval (vehicles carrying GPS or cellular devices that report their speed and location to a service provider that shares the data with the firm providing the NPMRDS data to U.S.DOT.). Understanding where the holes are in the available performance data (irrespective of the source) and deciding what to do about those holes are key tasks in the quality assurance task described in Section 4.6. Table 3-3 shows a sample of an NPMRDS. The NPMRDS is provided in two parts. The first part is a Traffic Message Channel (TMC) static file that contains TMC information that is updated only as necessary (see Table 3-3a). The second part is a database file set of average travel times (in seconds) of passenger, freight and combined for NPMRDS roadways geo-referenced to TMC location codes (see Table 3-3b). It includes travel speed measurements [collected 24 hours a day in 5-minute increments (epochs) when available] from GPS or cellular devices in the traffic stream. Other roadway performance data sources also can be used to provide estimates of vehicle travel times/speeds. Data sets similar (or in greater detail) to the NPMRDS are available from a number of private-sector firms, and these can be used in place of, or as a supplement to, NPMRDS. Data from agency-supported fixed sensors, such as roadway loops, also can be used to supplement the vehicle probe data. ITS (intelligent transportation systems) detectors are partic- ularly noteworthy because they have the capability to provide both speed and volume data (clas- sified by vehicle length) if managed appropriately. These data also can be used independently, or they can be combined with the NPMRDS to create a richer roadway performance data set. a. TMC Static File TMC Direction Admin_ Level_1 Admin_ Level_2 Admin_ Level_3 Distance (miles) Road Number Road Name Latitude Longitude 101N04099 Eastbound U.S. Illinois Cook 3.27285 I-90 Kennedy Ex 37.9615 -121.6961 101N04100 Westbound U.S. Illinois Lake 0.88324 I-290 Eisenhower 37.9906 -121.6972 b. Travel Time File TMC Date Epoch Combined Passenger Freight 101N04099 04022012 33 105 99 123 101P04099 04022012 78 98 92 125 101N04100 04022012 5 46 38 51 101N04100 04022012 31 45 39 52 Table 3-3. Sample NPMRDS data.

Identify and assemble Data 27 The key is that each state or MPO has access to data that allow the identification of travel speed-related bottlenecks. Using the NPMRDS—or other available data sets—states and MPOs can compute at a minimum when, where, and to what extent delays are occurring for both cars and trucks throughout the NHS. 3.3.2 Volume Data Volume data provide two things in a typical bottleneck analysis: 1. An estimate of the “usage” of a roadway because not all roadway segments are the same and 2. A way to perform weighted averages for index measures to produce facility, regional, or area- wide statistics. As with vehicle travel time and speed data, there are multiple sources of truck and traffic volume data that are available to state agencies and MPOs. Truck and traffic volume data can be obtained from the HPMS data that states submit to FHWA each year. The HPMS submittal describes AADT on each roadway segment of the NHS, as well as the percentage of trucks using those roadway segments. However, there are some challenges with using HPMS data as the source for truck classifica- tion data. The data tend to be 2 to 5 years old and based on a few days of classification counts. Much of the truck percentage data available on HPMS segments are actually estimates. The method for estimating truck percentages varies and can range from using truck percentages from counts nearby the segment to using truck percentages of roadways with similar functional clas- sification. This limits the accuracy of the count data in terms of calculating truck delay. HPMS data can be supplemented by using other sources that provide broader coverage over time and functional classification such as weigh-in-motion data and closed caption television. Ideally, the vehicle classification count data would have a much higher level of temporal reso- lution than average annual conditions. Thus, if HPMS data are the primary source for truck and traffic volumes, additional effort is needed to understand how traffic volumes vary over time at each roadway segment. State DOTs perform some level of short-duration truck and traffic volume counting, and these counts are frequently supplemented by continuously operating, permanent counters. Both of these data collection efforts provide volume estimates at a mini- mum hourly resolution. The combination of these data sources serves as the basis of the annual traffic estimates submitted in the HPMS (e.g., AADT and truck percentages). They also can be used to estimate the time-of-day traffic volume profiles present on roads. Other potential sources of time-of-day volume data (by vehicle classification) include daily volumes from a roadway inventory database and classification data from national-level sources such as FHWA. More information regarding each of these methods can be found in recently completed FHWA research. (11) 3.3.3 Other Data Sets To complete the data analysis structure needed for comprehensive bottleneck analysis, a vari- ety of other data sets will also be needed. As noted earlier in this section, these data include data on temporary operational capacity reductions that can cause delays to form, such as: • Vehicle crashes and other incidents. Most state DOTs have a safety branch that collects and makes available crash data. Figure 3-1 is an example of site to order crash data from the Iowa Department of Transportation. Another example is the multi-agency Regional Integrated Transportation Information System (RITIS), which allows participating agencies to access

28 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks information on incidents, including the types of vehicles involved and the timeline of the incident. • Construction activities. State DOTs track and announce construction and roadway closures and often store this information. Figure 3-2 is an example of an announcement from the Washington State DOT of work zone activity. • Bad weather. Historical weather data can be ordered from NOAA. Figure 3-3 is an example of an order for 15-minute precipitation data. • Special events. For process-based delays such as port gates, border crossings, intermodal railyards, weigh sta- tions, and toll plazas, facility-specific data sets are needed. For port gates, most terminal opera- tors maintain information on the dwell time of trucks within the port gates. Video cameras are typically needed to measure delay of vehicles waiting outside the port gates. In theory, truck GPS data can also be used to estimate the times that individual trucks spend waiting in line at Source: http://www.iowadot.gov/crashanalysis/index.htm. Figure 3-1. Crash database example from the Iowa Department of Transportation. Figure 3-2. Work zone log output from the Washington State DOT.

Identify and assemble Data 29 port gates and dwell times inside port gates. However, in practice, the level of geographic preci- sion needed to conduct this type of analysis makes the use of truck GPS data for these purposes challenging. Similarly, most border crossing facility operators maintain data that estimate delay approach- ing border crossing facilities along with border crossing time at the facilities. Truck GPS data can be used at these locations with appropriately located screenlines that allow for the measurement of time that passes between upstream and downstream locations from a border crossing facility. This process would provide information on a combined wait time and processing time at these facilities. Alternatively, roadside truck surveys can be used to collect information from truck drivers on their estimates for time spent waiting to travel to border crossing locations and time spent being processed at these facilities. Weigh stations feature two types of truck delay. There is the delay that occurs when traveling on the weigh-in-motion portion of the station when trucks are not asked to stop at the station. Then, there is the processing time for trucks that are stopped at the station. The delay on the weigh-in-motion portion of the station occurs on the approach to the weigh station where trucks must slow down even when there is no traffic or trucks may become queued at these locations when the volume of trucks exceeds the capacity at these locations. These speeds can be identified using the truck GPS methods mentioned throughout this chapter. There are not any standard- ized sets of data that measure the processing time for trucks that stop at weigh stations, but it could be estimated through either observation or roadside truck surveys. Speeds at toll plazas can also be estimated using truck GPS data. Source: http://www.ncdc.noaa.gov/cdo-web/search. Figure 3-3. Example of a weather data order from NOAA.

30 Guide for Identifying, Classifying, evaluating, and Mitigating truck Freight Bottlenecks These data sets can be quite large and contain many data items. Not all of the data items pre- sent in the base data systems should be brought into this analysis process. That is, not all vari- ables in the vehicle crash records are needed in the bottleneck analysis data structure. Finally, comprehensive truck bottleneck analysis requires information about the various types of physical disruptions that affect truck travel. These include: • Roadway geometric limitations (e.g., narrow lane widths, low-height bridges); • Grades steep enough to affect truck speeds; • Activities that delay vehicles (e.g., toll booths, weigh stations, international border crossings); and • A lack of truck-specific, last-mile facilities such as parking or load zones. Many of the physical disruption data elements are available through state DOT roadway inventory systems. Others will require independent research and data assembly activities. As with the other data sets, the availability of these data statewide (or regionwide) is not a require- ment for performing useful truck bottleneck analyses. However, the more of these data available for analysis, the more robust the outcome of the truck bottleneck analysis will become.