Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 22
22
may be an actual odometer reading read directly from the 3.4 Consideration of Temporal and
truck's equipment. It may also be a computed odometer Seasonal Impacts
reading based on the continuous GPS readings, in addition to
the recorded and reported readings. The truck-generated Freight flows are traditionally expressed as tons per year.
odometer reading was used in this research to compute truck This is true when freight flows are reported as multimodal
trip characteristics. This is because the GPS odometer read- flows (in CFS or FAF), as modal flows (e.g., in the STB Way-
ing will be incorrect when the device loses a satellite signal bill for railroads), or facility flows (as in the U.S. Army Corp
lock (e.g., if travelling in a tunnel or due to other sky block- of Engineers' Waterborne Commerce Statistics for ports).
ages). The difference between the average airline distance and Conversely, freight flows, particularly vehicle flows, are typi-
the actual highway mileage shown is an indication of the cally expressed in vehicles per day for capacity expansion and
amount of circuity of the trip on the highway network. design decisions, and as vehicles per hour to support opera-
The travel time was computed as the time difference tional decisions. In order to use existing freight and truck
between stop event times, adjusted to account for the dwell modeling processes to support infrastructure decisions, it
time at a stop. In order to calculate the dwell time at a stop, the would be useful to develop factors that can be used to convert
GPS data were examined to identify the first moving GPS annual freight flows to daily and hourly flows.
event record after a stop. The GPS vendor recorded this event Databases of truck vehicle movements are for specific geo-
when the vehicle speed after a stop reached 5 mph. This was graphic locations on highway networks. These databases of
considered to be a reasonable approximation of when the out- truck classification counts can not distinguish trucks by body
bound trip began. A flaw was discovered when it was deter- type or by the contents or type of freight being carried. The
mined that the recorded stop events could include both annual modal and multimodal commodity freight flows
business stops and stops captured due to congested traffic. For reported by trucks are for origins and destinations of the
the purposes of this research, only business stops were of freight flows, not for the highway locations along the routes
interest. Despite efforts to apply filters to exclude traffic stops, between origins and destinations that would correspond to
no consistently successful filter was identified. Based on this the truck counts.
feedback, the GPS vendor is intending to change the record- To address this difference, the research team first identified
ing logic to distinguish between stops with ignition on (a stop methods to assign commodity truck OD flows to the highway
in traffic) and a stop with the ignition off (an activity stop). network. This would allow the identification of freight flows by
This may be a better way to address this issue in the future. commodity at highway locations corresponding to the truck
The processed data for the four selected metropolitan areas counts. These truck counts could be used to develop monthly
are included in Table 3.2. Although not presented, these data and hourly factors. These factors could then be applied to com-
could be processed to determine additional information (e.g., modity flows at each specific location.
median, standard deviation, and distribution around the The flows by commodity will vary on the highway network,
median). Similarly, in addition to calculating average times and the monthly and hourly factors from counts will vary by
and distances between stops, the same information can be cal- location. However, if the commodity flow is principally of a
culated by stop sequence (1st, 2nd, 3rd, etc.). The ability to specific commodity, the variation in truck counts should also
develop this information may assist in developing chaining reflect variation for this commodity. For example, if the com-
and/or distribution models. modity truck flows at a location hypothetically consisted of
This research shows that subscription GPS data from trucks only a single commodity, then the monthly and hourly vehicle
may be an inexpensive way to determine a variety of character- counts at this location could be expected to represent the
istics that could be used in truck trip distribution and chaining. monthly and hourly factors for that commodity. Where no sin-
The data developed appear reasonably consistent and credible. gle commodity dominates, it is proposed that the truck flow
Before these data--or other data that would be developed in a pattern at any single location will reflect the seasonal and tem-
similar manner--could be fully utilized, questions regarding poral flow pattern of the underlying commodities. Although
data expansion need to be addressed. GPS subscription data this method cannot be expected to develop factors that would
are made anonymous before release in order to protect the apply to specific locations, the aggregation of the resulting pat-
identity of trucking clients. In order to develop meaningful dis- tern across all locations and commodities is expected to reflect
aggregations or expansions to types of trucks, more detailed an average national distribution of commodity freight flows by
information should be developed. Based on this research proj- month and by hour.
ect, the GPS vendor is investigating methods to store informa- For this research topic, the first step will be to develop a
tion (e.g., the first eight characters of a Vehicle Identification method to assign the truck commodity flows to the highway
Number [VIN], which could provide sufficient information to network. The second step will be to develop monthly and
develop disaggregations and survey expansion factors while hourly factors from national databases of truck counts. The
preserving anonymity). third step will be to apply those factors to the commodity
OCR for page 23
23
flows on the network corresponding to the count locations. ment), not the number of establishments should be used to
The fourth step will be to aggregate those monthly and hourly disaggregate freight flows.
flows and develop average national adjustment factors. For FHWA, Cambridge Systematics developed a procedure
that disaggregates FAF2 regional flows to county flows using
the employment data for the zones that produce and consume
Development of a Commodity Assignment
freight. This procedure relies on county business patterns'
The 1998 version of the Freight Analysis Framework (FAF1) employment, for each county, in the industries that produce
produced maps of truck flows on the FAF1 highway network and consume freight. The level of use by each industry for each
and made available (via download) highway network files of SCTG2 commodity in the FAF2 was established by regression.
daily freight truck volumes on the highway links, in two widely Additionally, flows through ports, airports, and border cross-
used platforms, ESRI and TransCAD. These flows were pro- ings were disaggregated based on the county in which the
duced by converting the county-to-county tonnage flows by facility was located and the share of reported activity at that
truck to daily truck flows through the use of annual-to-daily facility from other modal databases. This procedure is docu-
factors as well as tons-to-truck-payload factors. Although these mented in an unpublished FHWA report.10
data could have been used to produce highway flows by com- These disaggregation methods were used to convert the
modity, the proprietary nature of the FAF1 data prevented the 144-zone by 144-zone FAF2 flows (where the regional zones
disclosure of highway link flows with this information. It also beyond 114 represent ports, border crossings, and interna-
prevented the disclosure of the origin/destination/commodity tional zones) into flows among the 3,140 counties in the
(O/D/C) table at a county-to-county level, which could have United States. The FAF2 database is available for download
been used to assign the truck flows to the highway network. from the FHWA website as a Microsoft Access database. This
Only the reporting of state-to-state flows was publicly available. database was converted to a set of TransCAD matrices, one
The FAF2 flow database, which is the 2002 update to the for each SCTG2 commodity.
FAF1 flow database, can be considered as a basic O/D/C table The FAF2 technical documentation also describes a proce-
for 114 very aggregate zones, called FAF2 regional zones. Those dure for assigning truck flows to the FAF2 highway network.11
FAF2 zones consist of the state portion of the largest metropol- That procedure uses the stochastic user equilibrium (SUE)
itan areas, and the remainder of states or whole states outside assignment routine in TransCAD to assign the freight flow
of these metropolitan zones. FAF2 separately developed a high- tables. This assignment procedure uses the information avail-
way network, which included updated information and addi- able in the FAF2 network (such as capacity, total vehicle vol-
tional detail beyond the FAF1 highway network. However, the umes, and free-flow speed) to calculate a congested time on
regional zone structure of the OD table by commodity is not each link. That congested time is used as the basic link imped-
consistent with the assignment scripts for the FAF1 or the detail ance. That basic impedance is modified by additional infor-
of the FAF2 highway network. In order to be compatible with mation for each link, such as the number of lanes, the location
the assignment scripts developed for the FAF1, the O/D/C table of the link in urban areas, truck restrictions, truck route des-
must be disaggregated to smaller geographic zones (e.g., coun- ignations, tolls, and any interstate designation of the link. The
ties, as used in the FAF1). The FAF2 documentation9 describes SUE assignment is based on that modified impedance. The
a procedure for disaggregating the FAF2 regional zones to results of the assignment using the Battelle FAF2 disaggregated
counties and other freight activity centers. That procedure is database and this procedure is shown in Figure 3.1.
based on the share of the number of establishments in the For this research topic, TransCAD scripts were developed
activity center as a ratio of the number of establishments in the to implement the documented assignment procedure and
zone and the share of the HPMS truck VMT in the activity cen- were used to assign the county-to-county flows for each of the
ter as a ratio of the HPMS truck VMT in the zone. 42 SCTG2 commodities disaggregated from the FAF2 data-
There is no reason to think that truck trip ends in an activ- base. The resulting assignment of total truck tonnage is shown
ity center should be related to truck VMT in that activity cen- in Figure 3.2. The flow pattern appears similar to that of Fig-
ter. For example, a major truck route with considerable truck ure 3.1 where most flows are concentrated on major interstate
VMT passing through an otherwise empty activity center (e.g., highways and the trip ends are concentrated in the counties
county) does not indicate that there should be trip ends in that
activity center. Similarly, a measure of the level of intensity of
the establishments within an activity center (e.g., employ- 10 Cambridge Systematics, Inc., Development of a Computerized Method to Sub-
divide the FAF2 Regional Commodity OD Data to County-Level OD Data, prepared
for FHWA, January 2009, unpublished.
9 Battelle Memorial Institute, "Chapter 4: FAF2 Truck O-D Data Disaggregation," 11 Battelle Memorial Institute, "Chapter 5: Freight Truck Assignment and Cali-
in FAF2 Freight Traffic Analysis, FHWA, June 27, 2007, http://ops.fhwa.dot.gov/ bration," in FAF2 Freight Traffic Analysis, Federal Highway Administration, June
freight/freight_analysis/faf/faf2_reports/reports7/c4_data.htm. 27, 2007.
OCR for page 24
24
Source: FAF2 Freight Traffic Analysis, Figure 3.3.
Figure 3.1. Base-year 2002 FAF2 truck flow on FAF2 highway network.
with large populations or production and/or consuming tances of those links. The result of this calculation is shown in
industries. Table 3.4. As shown, the total ton-miles calculated for trucks
This assignment procedure provides link volumes for each are approximately twice the reported value in the CFS. This
of the 42 SCTG2 commodity flows that are not publicly avail- is to be expected since the CFS has several major commodity
able for the FAF2 network. Although the assignment routine gaps, referred to by the FAF2 as out-of-scope commodities.
can produce flows for each of the 42 disaggregated SCTG2 In addition, the CFS undercounts some categories of trade
commodities, there are known errors in the creation of flows and movements of freight (e.g., in-transit movements, petro-
for certain import and export flows. For example, in translat- leum products, and exports). The FAF2 includes these addi-
ing from the Performance Monitoring System (PMS) com- tional flows.
modity classifications used in waterborne commerce to the This summary of ton-miles indicates that the flows on the
SCTG2 commodity classification system used in FAF2, all FAF2 network are reasonable and can be used in processing
manufactured goods were reported to move in SCTG 34, the later steps.
machinery. As shown in Figure 3.3, this results in higher than
expected flows to and from ports such as Savannah, Georgia,
Development of Monthly
and Charleston, South Carolina.
and Hourly Truck Factors
This error is more pronounced at the SCTG2 level of detail
and is less of an issue at higher levels of commodity aggrega- FHWA maintains the VTRIS database of traffic counts
tion. Additionally, the level of detail for 42 SCTG commodi- taken at stations by automatic traffic recorders (ATRs), vehi-
ties has additional processing and reporting issues. Rather cle classification counters, weight-in-motion equipment, and
than processing the flows at the SCTG2 level, the flows were weight enforcement stations as submitted by state DOTs. The
grouped to the nine classes of commodities used in the 2002 VTRIS database includes the station description, vehicle clas-
CFS. For these CFS commodity groups, the ton-miles of flow sification, and the time of the counts in a consistent format for
were calculated from the assigned link volumes and the dis- all 50 states. Although this does not provide any information
OCR for page 25
25
Figure 3.2. Disaggregated FAF2 truck ton flows (all commodities).
about the commodities carried by these trucks, it is possible hourly factors could be developed for 623 links on the FAF
to develop hourly and monthly factors from the stations in highway network. The location of these stations is shown in
VTRIS. Weigh-in-motion and weight information stations do Figure 3.4.
not provide the continuous readings that would be required To develop monthly allocation factors, a station needs to
to develop monthly and hourly factors. The ATR counts are of be operated without daily gaps for the year. There were only
total vehicles and do not differentiate between trucks and 200 stations without gaps that could be used to develop
other vehicles, including automobiles. The vehicle classifica- monthly factors. Only stations that match the FAF2 network
tion counts do provide the ability to distinguish trucks from links can be applied to the commodity flows and multiple sta-
other vehicles and include locations that are counted contin- tions on the same FAF2 link have to be combined before they
uously. The information in VTRIS for classification counts is can be used. Monthly factors can be developed for 177 links
recorded by hour and by date. The complete VTRIS database on the FAF2 highway network. The location of these stations
for 2007 was obtained from FHWA's Office of Highway Pol- is shown in Figure 3.5.
icy Information. From that VTRIS database, the tables of vehi- For each station, factors were developed for combination
cle classification counts were selected. trucks--Vehicle Classes 8 through 13--according to FHWA's
VTRIS contains records for 13,862 classification stations for Scheme F classification. These vehicles are those that would
the United States. However, in order to develop hourly allo- most likely carry freight.
cation factors, a station needs to be operated without hourly
gaps for weekdays. There were only 798 stations without gaps,
Apply Factors to the Commodity Flows
and these could be used to develop hourly factors. Only sta-
on the Network
tions that are on the FAF2 network links can be applied to the
commodity flows and multiple stations on the same FAF2 link For each of the 623 links that have complete hourly factors
have to be combined before they can be used. As a result, developed from counts, those factors were applied to the
OCR for page 26
26
Figure 3.3. Disaggregated FAF2 truck ton flows (SCTG 34, machinery).
Table 3.4. Annual ton-miles traveled by each CFS commodity group.
CFS
Commodity SCTG Annual Ton-Miles
Group Code Description (Millions)
CFS1 01 to 05 Agriculture Products and Fish 398,154
CFS2 06 to 09 Grains, Alcohol, and Tobacco Products 210,034
CFS3 10 to 14 Stones, Nonmetallic Minerals, and Metallic Ores 382,451
CFS4 15 to 19 Coal and Petroleum Products 222,452
CFS5 20 to 24 Pharmaceutical and Chemical Products 268,038
CFS6 25 to 30 Logs, Wood Products, and Textile and Leather 330,683
CFS7 31 to 34 Base Metal and Machinery 437,008
CFS8 35 to 38 Electronic, Motorized Vehicles, and Precision Instruments 87,758
CFS9 39 to 43 Furniture, Mixed Freight, and Miscellaneous 245,299
Manufactured Products
Total 2,581,876
Truck Ton-Miles--2002 CFS Table 2a 1,261,813
OCR for page 27
27
Figure 3.4. VTRIS stations with valid hourly factors.
Figure 3.5. VTRIS stations with valid monthly factors.
OCR for page 28
28
annual tonnage flows to produce estimated hourly flows for significantly from national averages, absent any other local
each of the SCTG commodities. For each of the 177 links that information, it is reasonable to assume that commodity groups
have complete monthly factors developed from counts, those all follow the same distribution pattern for both hourly and
factors were applied to the annual tonnage flows to produce monthly flow. Although this pattern could be true for large
estimated monthly flows for each of the SCTG commodities. commodity groups, it might not hold true for the very differ-
ent SCTG2 commodities with these groups. The pharmaceuti-
Develop Average National cal and chemical commodity group includes two commodities
Adjustment Factors (SCTG 21, pharmaceuticals, and SCTG 22, fertilizers) that
could be expected to follow very different patterns. However,
The hourly flows for each location were used to develop a the development of hourly factors from the commodity flow
national hourly summary of freight flows. The resulting hourly data for each of these commodities produces very similar
freight flows were aggregated to the nine CFS commodity results, as shown in Figure 3.8. It is therefore assumed that the
groups. An hourly distribution of the summary of freight flows method used to estimate the average hourly distribution of
was developed and that distribution is shown in Figure 3.6. It commodities would yield the same results for any commodity.
is noted that averaged over all locations, the hourly distribu- Monthly flow patterns should be related to monthly pat-
tion of each of the nine CFS commodity groups appears to be terns of production by commodity. This information can be
virtually identical. verified from separate sources. The Federal Reserve Board
The monthly flows for each location were used to develop tracks this information in its industrial production and capac-
a national monthly summary of freight flows. The resulting ity utilization statistics. The U.S. Census Bureau tracks this
monthly freight flows were aggregated to the nine CFS com- information in the Manufacturers' Shipments, Inventories,
modity groups. A monthly distribution of the summary of and Orders Survey (M3). The data were examined for 2007,
freight flows was developed and that distribution is shown in which was prior to the current economic recession. Both data
Figure 3.7. It is noted that averaged over all locations, the sources show similar results, but the Census Bureau data is
monthly distribution of each of the nine CFS commodity easier to use since it tracks data for the entire year, not by fis-
groups appears to be virtually identical. cal quarter. The Census Bureau survey reports shipments as
As noted, for both hourly and monthly factors, the resulting dollar values not tonnage, but if there is stability in commod-
distribution showed little variation by commodity group. This ity prices, the flow patterns for tons and value should be sim-
finding was not expected. It seems to suggest that while indi- ilar. Additionally, the Census Bureau reports these shipments
vidual locations could show distribution patterns that differ by NAICS industry, which is similar, but not identical, to the
7.0%
6.0%
CFS1
5.0% CFS2
CFS3
4.0%
%Hourly
CFS4
CFS5
3.0% CFS6
CFS7
2.0% CFS8
CFS9
1.0%
0.0%
AM
AM
PM
PM
AM
AM
AM
AM
PM
PM
PM
PM
12
10
12
10
2
4
6
8
2
4
6
8
Hour Beginning
Figure 3.6. Hourly distribution of truck commodity flows.
OCR for page 29
29
Figure 3.7. Monthly distribution of truck commodity flows.
SCTG commodities in the FAF. Table 3.5 shows the reported monthly shipments is shown in Figure 3.9. Similar to the esti-
value of shipments by industry by months as well as the cal- mated monthly distribution of flows by commodity shown in
culated standard deviation of those monthly flows as a per- Figure 3.7, there is little variation of flows across the months.
centage of average monthly flows. For all but a few industries, The industries, whose monthly standard deviation of flows
which are shown on five shaded rows, monthly variation as exceeds 10 percent of the average, are shown in Figure 3.10.
a standard deviation is less than 10 percent of the average Even for most of the industries with the most variation, that
monthly flow. variation is minimal. For those industries with the most vari-
For the industries excluding those in Table 3.5's shaded ation, since the reported data is shipment value not tons, it is
rows, the distribution of the U.S. Census Bureau reported conceivable that the variation is due to fluctuations in com-
Percent
7%
6%
5%
4%
3%
2%
1%
0%
12 A.M. 2 A.M. 4 A.M. 6 A.M. 8 A.M. 10 A.M. 12 P.M. 2 P.M. 4 P.M. 6 P.M. 8 P.M. 10 P.M.
Fertilizers Pharmaceuticals
Figure 3.8. Hourly distribution of pharmaceutical and fertilizer flows.
OCR for page 30
Table 3.5. Value of shipments by industry (2007, in billions of dollars).
Standard
Deviation as
Percentage of
Industry January February March April May June July August September October November December Average
Food Products 44.1 44.3 47.1 44.7 47.7 47.9 46.3 50.2 49.9 51.3 50.6 49.4 5%
Beverage and Tobacco 9.4 9.6 10.7 10.6 12.1 11.8 11.2 12.3 10.9 11.7 11.4 10.6 8%
Products
Textiles 3.0 3.1 3.2 3.1 3.2 3.2 2.9 3.2 3.1 3.1 2.9 2.6 6%
Textile Products 2.4 2.5 2.7 2.6 2.6 2.8 2.6 2.7 2.5 2.6 2.4 2.1 7%
Apparel 2.2 2.5 2.5 2.3 2.4 2.3 2.4 2.8 2.6 2.8 2.9 2.3 9%
Leather and Allied Products 0.4 0.6 0.6 0.5 0.5 0.5 0.5 0.6 0.6 0.5 0.5 0.5 10%
Wood Products 7.3 7.5 8.4 8.6 9.3 9.6 8.8 9.4 9.0 8.8 7.8 7.4 10%
Paper Products 14.0 13.2 14.1 13.6 14.4 14.5 14.0 14.5 14.0 14.5 14.0 13.8 3%
Printing 8.0 7.8 8.7 8.2 8.3 8.4 8.0 8.7 8.7 9.4 9.1 8.5 6%
Petroleum and Coal Products 36.6 36.4 43.1 45.2 50.4 49.1 49.9 47.8 47.7 50.2 54.7 53.0 12%
Chemical Products 51.8 50.8 58.0 55.9 58.2 57.1 55.2 56.9 53.7 58.0 54.3 54.3 4%
Plastics and Rubber Products 16.5 16.0 18.1 17.7 18.9 18.6 17.4 18.9 17.3 18.8 17.1 15.3 7%
Nonmetallic Mineral Products 9.1 9.0 10.4 10.4 11.0 10.7 10.2 10.9 9.8 10.6 9.4 7.9 9%
Primary Metals 19.5 18.8 21.0 20.8 21.9 21.3 19.5 20.9 19.9 21.3 19.1 17.5 6%
Fabricated Metal Products 25.3 25.3 28.0 27.3 28.8 28.9 26.7 30.0 27.7 28.6 25.9 24.1 7%
Machinery 23.2 24.6 30.1 29.0 28.9 30.1 26.8 27.7 28.5 28.2 25.9 28.5 8%
Computer and Electronic 28.4 29.6 35.7 30.1 30.4 36.3 27.4 31.0 35.9 31.5 32.4 38.1 11%
Products
Electronic Equipment, 9.3 9.6 11.3 10.6 10.9 11.3 10.0 11.2 11.3 11.0 10.3 10.1 7%
Appliances, and Components
Transportation Equipment 50.3 55.9 65.9 55.2 61.9 63.7 45.3 64.5 58.5 60.6 57.2 54.0 11%
Furniture and Related Products 6.5 6.7 7.1 6.7 6.8 7.0 6.8 7.4 6.9 7.1 6.7 6.5 4%
Miscellaneous Products 11.7 11.7 13.6 11.9 12.6 13.6 11.6 12.9 12.9 13.4 13.2 13.9 7%
Source: U.S. Census Bureau, Manufacturers' Shipments, Inventories, and Orders Survey (M3).
Note: Shaded rows indicate those industries with a standard deviation of 10 percent or greater.
OCR for page 31
31
Dollars (in Millions)
$350,000
$300,000
$250,000
$200,000
$150,000
$100,000
$50,000
$0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2007
Figure 3.9. Industries with minimal monthly variation.
modity price (e.g., the value per ton of petroleum) or in inter- that have been converted from annual to daily flows based on
national currencies. The U.S. Census Bureau data generally averages need not be concerned about seasonal variations in
confirms the findings of the proposed method, that there is lit- those commodities.
tle variation in commodity flows, on average, throughout the Similarly, when policy decisions need to consider the hourly
year. In the absence of local data showing specific local varia- variation of commodity flows, absent any specific local infor-
tions, any policy considerations of commodity truck flows mation, the hourly distribution of commodity flows according
Dollars (in Millions)
$70,000
$60,000
$50,000
$40,000
$30,000
$20,000
$10,000
$0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2007
Petroleum and Coal Products Computer and Electronic Products
Transportation Equipment Leather and Allied Products
Wood Products
Figure 3.10. Industries with the most monthly variation.
