2.1  Introduction

This chapter describes the methodology of building a freight planning network by utilizing the 2002 HPMS database for the subsequent freight assignment modeling task. An exclusive freight network was developed to support a comprehensive freight analysis. This analysis in turn was used to identify system-wide capacity deficiency elements for the nation’s highway and to synthesize freight movement related to capacity issues. The network is also intended to support FHWA freight policy tools. These tools help FHWA assess capacity deficiencies of the freight transportation system and identify specific needs. One important task of the freight network development was to integrate the preliminary version of the FAF² network with 2002 HPMS data elements essential to simulate and subsequently forecast the impacts of truck flows on the nation’s highway system.

2.2  NHPN Version 2005.10 – The Source Network

The geospatial coverage of the FAF² network was developed using FHWA NHPN Version 2005.10. The NHPN is a comprehensive geographic information system (GIS)-based network database of the nation’s major highway system. It represents more than 450,000 miles of the nation’s highways comprised of Rural Arterials, Urban Principal Arterials, and all National Highway System (NHS) routes. (In 1997, the U.S. public road system included 433,311 miles on the arterial system including 46,068 miles on the Interstate system alone.)  Out of 454,662 miles of the NHPN network, 452,254 miles are included in the FAF². The following roadways are included:

• Interstate highways
• NHS links
• National Network that are not part of NHS
• Other rural and urban principal arterials
• Intermodal connectors
• Rural minor arterials for those counties that are not served by either National Network NN or NHS
• Urban streets as appropriate for network connectivity.

In recent years, Battelle worked with the Office of Intermodal & State Programs, FHWA, to integrate the HPMS linear referencing geospatial schema as part of the NHPN network to allow capturing HPMS data elements. The HPMS is the nation’s highway database maintained by the FHWA using data supplied by the states and updated on a regular basis. States are required to report certain information for every segment of public highway and roadway. For example, the states must report mileage, average annual daily traffic (AADT), route number, jurisdiction, functional classification, number of lanes, service flow ratio (also a measure of capacity), and pavement condition. In addition, the state must report supplementary information for statistically valid samples of roadway sections selected on the basis of functional classification and volume group. These particular highway sections are called “sample segments,” as opposed to the former segments referred to as “universe segments.”  The additional data required for the sample segments include detailed pavement information, geometric data, truck percentage, traffic/capacity data, and environmental data.

The original NHPN linear referencing system (LRS) architecture was based on a non-geometric-based spatial layer that only allows HPMS data visualization through the dynamic segmentation technique, where each NHPN link is transparently segmented by using the respective HPMS data section (a section is normally defined by the smallest denominator of the attribute data). This creates difficulties for network analysis, because a network is defined by a fixed set of nodes and links. To overcome such limitation, the NHPN version 2005.10 accommodated several changes so that HPMS data can be transferred or aggregated to an equivalent NHPN link. The changes include, but are not limited to, the following:

Assign HPMS LRS route information to each set of NHPN links that comprise the HPMS route. The route information includes unique LRS identification and beginning and ending anchor mileposts.

Modify NHPN topological link direction to state-specific HPMS LRS topological direction.

Assign a sequential number for each set of NHPN links that comprise a unique HPMS LRS with beginning and ending milepost values for each NHPN link.

Calibrate NHPN milepost values for each NHPN link with HPMS LRS milepost anchor values. This is done to compensate the measurement error due to NHPN map scale (error between a ground measurement and map resolution).

Table 2.1 illustrates how one unique HPMS LRS “000000090E00063” is spread out over ten NHPN links, where each link has been assigned a milepost value and a sequence number along an HPMS LRS topological direction (defined in the next section). This data structure makes it easier to run SQL for integrating HPMS and other state data that reside under LRS schema.

Table 2.1: Relationship Between HPMS LRS Milepost and NHPN Node

NHPN Link ID	HPMS Unique LRS	Unique LRS Beginning Milepost	Unique LRS Ending Milepost	NHPN-Node Equivalent Milepost (start)	NHPN-Node Equivalent Milepost (end)	NHPN Link Sequence
162266	000000090E00063	0.000	20.280	0.000	1.304	1
155342	000000090E00063	0.000	20.280	1.380	2.674	2
155346	000000090E00063	0.000	20.280	2.830	3.014	3
155344	000000090E00063	0.000	20.280	3.190	3.373	4
155340	000000090E00063	0.000	20.280	3.570	4.169	5
155338	000000090E00063	0.000	20.280	4.410	4.499	6
155336	000000090E00063	0.000	20.280	4.740	4.839	7
155333	000000090E00063	0.000	20.280	5.080	5.770	8
153471	000000090E00063	0.000	20.280	6.010	7.602	9
153472	000000090E00063	0.000	20.280	7.840	20.280	10

A line layer (or network) in a map consists of many line features, each of which begins and ends at a point called an “endpoint,” and each of which is defined by a series of coordinates. Each line feature has two directions. These are the topological direction and the flow direction. The topological direction of a line feature is defined by the order in which the coordinates appear, and it has little or nothing to do with direction in which trucks can travel. If the beginning coordinate of a line feature is A and the ending coordinate is B, the topological direction of the line is AB and the opposite direction is BA. However, both “topological” and “flow direction” are important features for a network to be used in freight assignment procedures.

The “topological” direction ensures quality integration of HPMS data that are collected in some fashion of topological or survey chain direction. Typically, most states collect the HPMS data for a given highway from south to north and from west to east. For network analysis, “flow direction” ensures that flow is assigned to one-way pairs according to its flow direction. Another important component of network development is node and link connectivity.

FAF² network development utilizes both TransCAD and ArcGIS software programs with custom macros to ensure proper topological and flow direction including network connectivity. For flow direction, appropriate one-way pairs were added to links for those cities with population greater than or equal to 100,000. In addition, highway links with posted prohibitions for truck traffic were also flagged. Both truck restriction and flow direction information was obtained from the Rand McNally 2006 Motor Carrier Road Atlas, Microsoft MapPoint, and state DOT websites. Figures 2.1 and 2.2 illustrate the network topology and flow direction respectively.

 

Figure 2.1:  HPMS Data Chain or Network Topological Direction

 

Figure 2.2:  Network Flow Direction “One-Way” Pairs

 

The preliminary FAF² network coverage provided to Battelle by FHWA does not include a roadway connection to Alaska. The connection is important to allow highway freights that move between the lower 48 states and Alaska via Canada. The connection between Alaska and the lower 48 states was accomplished using the Alaska Highway and Canadian highways. This enhancement allows FAF² freight to be assigned the continental U.S. and Alaska. For the Canadian portion of the network, Battelle used the routes taken by trucks based on QUALCOMM GPS route information. The QUALCOMM route information was obtained from a recent study by Battelle for the FHWA Office of Motor Carrier (Expanded Satellite-Based Mobile Communications Tracking System). Figure 2.3 illustrates the Alaska connection to Mainland USA via Canada.

 

Figure 2.3:  Alaska Highway Connection to Mainland USA via Canada

 

2.3  Linking HPMS Data to the FAF² Highway Network

The FAF² Highway network contains HPMS 2002 route information through which HPMS traffic and other network modeling data are transferred or aggregated. Since HPMS uses sample sections to report highway performance-related data items, procedures were developed to capture attributes from a sample section and then spread these attributes to FAF² geospatial network links. The transformation of HPMS tabular data to the FAF² geospatial network was accomplished in five steps, as listed below.

2.3.1  Update FAF² LRS Schema with HPMS LRS Data Schema

The preliminary version of the FAF² network contains HPMS 2002 route information through which HPMS traffic and other assignment-specific data (traffic, truck percentage, lane, capacity, and speed) are linked to FAF². The quality of data transformation from HPMS to FAF² network segments depends upon the accuracy of LRS and milepost anchor values. Depending on the states, the current HPMS LRS matching rates with the preliminary FAF² segment-specific matching rate is between 71.5 and 100 percent (www.nhpnlrs.org). Since FAF² network “LRSKEY” and milepost anchor values are current only to the HPMS 2002 data schema, additional work was conducted to verify or improve the LRSKEY matching rate with HPMS 2002. For this task, Battelle utilized the state GIS-based LRS submittal currently maintained by Battelle as part of the FHWA NHPN maintenance and development project. This subtask produced an updated FAF² network with HPMS 2002 “LRSKEY” and milepost anchor values.

2.3.2  Load HPMS Traffic and Roadway Performance Data

The HPMS database was loaded onto FAF² using the “LRSKEY” field attributes of FAF² network segments and the HPMS LRSID attribute. The LRSID field, along with beginning and ending milepost information and route sequential direction, was used to link the FAF² with the HPMS. As illustrated in Table 2.1, even though the same LRS can exist in both FAF² and HMPS, these two datasets could not be linked directly because the milepost values are not exactly the same. For example, for a given LRSKEY, FAF² may have one link that represents
0 to 20 miles. For the same FAF² link, the HPMS may have two or many universe or sample segments for this 20 mile long FAF² segment. To transfer the required HPMS attributes, e.g., AADT to FAF², first, all HMPS records were identified that belong to this FAF² link using a predefined tolerance, for example a half mile, to match data from these two datasets. After identifying the most appropriate overlay, HPMS sections an average value for the AADT, which is calculated and assigned to the FAF² link.

HPMS reports two sets of data:  (1) universe data—where data elements are available for most of the FAF² network; and (2) sample data—where data elements are available for randomly selected highway sections.

The loading of HPMS data onto the FAF² network was achieved in two steps. First, the HPMS data for both the sample section and the universe section were loaded onto the FAF² link. This was done by averaging all of the smallest units of the HPMS data elements that belong to a FAF link as appropriate to merge the FAF² network. Some data elements, for example, “Terrain,” are based on the qualitative assessment of the HPMS link and do not allow for averages. For these types of data elements, the values were directly carried over using a predefined condition. Table 2.2 shows applied conditions for the HPMS data elements.

Table 2.2:  HPMS Data Elements Used in the Study

HPMS Item No.	Variable Description	Methodology	Condition
HPMS Universe Data Elements
9	Rural/Urban Designation	Carry Over	Max (of Codes)
14	National Highway System	Carry Over	Max (of Codes)
23	Designated Truck Route	Carry Over	Max (of Codes)
24	Toll	Carry Over	Max (of Codes)
28	AADT	Average
30	Number of Lanes	Carry Over	Min (of Values)
HPMS Sample Data Elements
59	Type of Terrain	Carry Over	Max (of Codes)
63	Speed Limit	Carry Over	Min (of Values)
65A	Percent Single Unit Trucks	Average
65B	Percent Combination Trucks	Average
66	K-Factor	Average
68	Peak Capacity	Average
69	Volume/Service Flow Ratio	Average
73	Future AADT	Average
74	Year of Future AADT	Carry Over

 

The HPMS data capture methodology can be described in sequential order as follows:

1. If a non-spatial record of HPMS is longer than the FAF² link-specific spatial records and the value is missing from the spatial link, then transfer attributes from the HPMS non-spatial link to the FAF² link. For example, if a non-spatial link has a milepost value from 0 to 10 and the spatial link has milepost from 5 to 7, then this FAF² link falls 100 percent within the HPMS section limit and the data are transferred directly.
2. For remaining non-matching FAF² link records, set the SQL condition for HPMS section-specific matching milepost tolerance to zero.
3. For HPMS sections shorter than the FAF² link, find all HPMS links that can be associated to a particular FAF² link. For example if a FAF² link is identified as route “A” and has milepost values of 0 to 10 miles, this step selects all the HPMS links that fall on route “A” with intersecting HPMS milepost.
4. Transfer data according to a user-selected method listed in Table 2. For a value that requires “carryover,” the value in the longest section of the selected HPMS dataset is used.
5. For the carry over method, an error is reported if there exist two or more different values for a group. For other methods, an error is reported when the value of selected HPMS records has a difference of more than 1.5 standard deviations from the average of the group. These errors are manually inspected at the end of the data capture.
6. For the next iteration, the HPMS milepost tolerance is increased by 0.1 miles.
7. Steps 3 to 6 are repeated until the maximum tolerance of one (1) mile is reached.

Figure 2.4 illustrates the data capture procedure from HPMS to the FAF² network.

2.3.3 Spreading HPMS Section Data to the FAF² Network Link

The data capture methodology applied in Section 3.2 is identical for both the universe and sample sections if an LRS matching route is available for both the FAF² link and the HPMS sections. Based on Battelle’s experience with the similar FHWA FAF¹, SMA, and NHPN projects, many FAF links will still exist with missing data for the transferred attributes. These links need to be populated by executing various spatial algorithms. For example, AADT for a missing link can be populated by taking the average of two adjacent links, or AADT data can be transferred from adjacent links if it is not part of an interchange.

 

Figure 2.4:  Data Capture Procedure

 

After merging the converted state-provided traffic data to the network, it was necessary to perform further GIS programming within the TransCAD software environment to fill gaps in the network for which there are no data. These gaps occur because each state route does not have HPMS data elements, plus some route information is incomplete in the original NHPN Version 2006.10 network. An example of data spreading for AADT or AADTT logic is described in Figure 2.5 below. Gaps in traffic data on the links are marked X. For (a), the value for X is the average on the two adjacent links i.e., (A + B)/2; for (b), X = B i.e., the value for the section immediately preceding it; and for (c) X = A i.e., the value for the section immediately preceding it before the branch-off link.

 

Figure 2.5:  Aggregation of Traffic Data

 

2.3.4  Assigning Data for Non-HPMS FAF² Network Links

Not every FAF² link has HPMS route information. Approximately 15 percent of the links in the FAF² network do not have any associated HPMS LRSKEY. A lookup table that lists critical HPMS attributes for assignment purposes (average AADT, K-factor, D-Factor, AADTT, Lane, etc.) and values for each functional class or NHS for the given county was developed by grouping the HPMS data appropriately. This lookup table was then used to populate the FAF² links that did not have any corresponding HPMS data or that could not be derived using spatial allocation algorithms. Table 2.3 lists the various lookup tables developed to populate missing data.

2.3.5  Quality Assurance Analysis

After completion of database loading, visual inspection was conducted by developing scale-based theme maps using the TransCAD GIS software. This approach is very effective to identify the major discrepancies between adjacent links (i.e., significant drop of AADT between two adjacent highway links) or among various functional classes. This approach also helps to identify the inherent anomalies with the HPMS database if any. The purpose of the quality assurance subtask was to manually check the accuracy of merged data. For example, if the difference in traffic value from one link to the next were greater than 20 percent, the original state traffic data collected in the FAF¹ project were consulted to verify if the accurate value had been merged to the network. If the values compared well with the state data, then the more common value for that link was used to ensure continuity in the traffic volume. These abrupt changes could also result from the merging process where aggregation was used. This is a smoothing process that served as a reasonableness check of the traffic data merged with the network.

Table 2.3:  FAF² Network Missing Data Lookup Tables

Lookup Table Name	Purpose
AADTLookup	National average AADT grouped by state and county codes and functional classification. This table is used to populate missing AADT of the network table.
AADTLookUpPerState	National average AADT grouped by state codes and functional classification. This table is used to populate missing AADT of the network table where the previous lookup table failed to do.
DFactorLookUp	For missing network D-factor values.
FixFClassLookUp	For missing functional classifications.
FreeWayFidLookUp	Adjustment factor for freeway interchange density.
FreeWayFlcLookUp	Adjustment factor for right shoulder lateral clearance; applies to freeway.
FreeWayFlwLookUp	Adjustment factor for freeway lane width.
FreeWayFnLookUp	Adjustment factor for number of lanes.
FreeWayRuralEtLookUp	Rural equivalence table for SU, Comb, ST, DT, and TT.
FreeWayTempEtLookUp	Equivalence table truck. This table is used only for peak hr factor (PHF) calculation.
FreeWayUrbanEtLookUp	Urban equivalence table for SU, Comb, ST, DT, and TT.
KFactorLookUp	For missing network K-factor values.
MultiLaneFaLookUp	For multi-lane adjustment factor for access points.
SpeedLookUp	For missing network speed values.
TruckLookUP	For missing non-network truck, SU and Comb.

 

2.4  Summary of Freight Network Development

The outcome of this task was a routable FAF² highway network loaded with 2002 traffic volume and other HPMS-attributable information required for the development of subsequent link parameters that are themselves required for freight assignment. Figure 2.6 illustrates the truck volume on the FAF² network developed using the HPMS 2002 database.

 

Figure 2.6:  HPMS 2002 Truck Volume on FAF² Network

 

 

previous | next