Professional Documents
Culture Documents
Abstract: This paper focuses on analyzing traffic facilities for an intermediate time frame. There are two methodologies examined in
this work, the first uses extrapolated, historical traffic count data and the second uses an urban transportation model. Using several
intersections within Huntsville, Ala., as case study intersection locations, this paper applies both methodologies to forecast near-future
traffic and compares the forecasted results with the actual traffic counts to determine which methodology better replicated actual volumes.
The results of this work indicate that a properly validated and applied urban transportation planning model has the ability to provide more
accurate traffic forecasts to support the traffic engineering analysis decision than the commonly used extrapolated traffic trends.
DOI: 10.1061/共ASCE兲0733-9488共2004兲130:3共159兲
CE Database subject headings: Traffic models; Traffic engineering; Forecasting; Traffic analysis; Urban planning.
Introduction and Background Barnett, personal communication, February 20, 2002兲. Trend line
analysis is commonly used to obtain future intermediate range
The analysis of highway transportation infrastructure generally traffic forecasts to be entered into traffic engineering analysis
uses existing traffic volumes or long-term traffic forecasts software. This allows the transportation analyst to develop a fore-
共Dickey 1983兲. The existing traffic volumes are used to analyze casted delay and level of service, based on existing roadway con-
short-term transportation improvements, focusing on projects to ditions.
handle specific problem locations and are typically justified using The use of trend line analysis for intermediate transportation
specific traffic engineering analysis software and techniques. analysis is limited in two significant ways. First, trend line analy-
Long-term traffic forecasts use projected socioeconomic data to sis does not respond to changes in the socioeconomic character-
justify major roadway infrastructure improvements, generally istics in the study area. For example, an area experiencing stable
using urban planning models running the standard four-step, growth will have stable traffic counts. However, a major change
travel-demand, and forecasting process. in the socioeconomic characteristics for an area can abruptly
Currently, there is no agreed-upon methodology for the inter- change the traffic conditions. This abrupt change in traffic condi-
mediate term of 3–7 years for traffic analysis. Typically, interme- tions cannot be modeled using trend line analysis and result in the
diate analysis uses traffic engineering analysis software with traf- need for postdevelopment roadway improvements, which are
fic and turning movement data developed through trend line often expensive and difficult. The second limitation of trend line
analyses 共Timothy Barnett, personal communication, February analysis is that traffic volumes are self-limiting. Implying that
20, 2002兲. Trend line analysis is a method to forecast future traffic traffic growth will not continue along a linear trend, especially as
based on historical data by developing a best-fit equation relating the volume nears capacity. A typical traffic growth curve contains
traffic volumes from different years to develop growth patterns. a linear portion where the trend line is applicable; however, as-
These trend lines can be of different types, such as linear, expo- suming linear traffic growth on the nonlinear sections of the
nential, power, logarithmic, and polynomial; however, linear is growth curve will lead to errors in the analysis.
the most common for intermediate traffic forecasting 共Timothy The goal of this paper is to examine the possibility of using
traffic data from a validated urban transportation model to calcu-
1
Assistant Professor, Dept. of Civil Engineering, Univ. of Alabama in
late forecasted delays and levels of service to evaluate intermedi-
Huntsville, Huntsville, AL 35899. E-mail: mikea@cee.uah.edu ate improvements in transportation supply. This paper presents a
2
Assistant Professor, Dept. of Industrial and Systems Engineering and study undertaken that developed the forecasted level of service
Engineering Management, Univ. of Alabama in Huntsville, Huntsville, 共LOS兲 for several intersections using both the linear forecasting
AL 35899. E-mail: gholston@ise.uah.edu of historical traffic data and a 1996 validated travel model with a
3
Graduate Student in Civil Engineering, Univ. of Alabama in Hunts- 5-year socioeconomic forecast. These values were compared with
ville, Huntsville, AL 35899. E-mail: akkinev@email.uah.edu actual counts, delays, and levels of service calculated in 2001.
Note. Discussion open until February 1, 2005. Separate discussions
must be submitted for individual papers. To extend the closing date by
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this technical note was submitted for review
Methodology
and possible publication on August 22, 2002; approved on April 25,
2003. This technical note is part of the Journal of Urban Planning and This section presents the work performed to test whether urban
Development, Vol. 130, No. 3, September 1, 2004. ©ASCE, ISSN 0733- planning models are capable of providing intermediate traffic
9488/2004/3-159–162/$18.00. counts for transportation analysis. Specifically, this section con-
Whitesburg
Airport and 43 132 86 12 189 89 107 89 10 103 118 72
Whitesburg
Planning Organization兲 共James Moore, personal communication, LOS for the trend line and the actual data do not agree for any of
April 22, 2002兲. The 1996 model was developed in that year and the case study intersections and for five of the six intersections,
validated to 1995 traffic counts for Huntsville. The original, vali- the trend line developed LOS was one letter grade higher than the
dated model was accompanied by a projected socioeconomic actual. One possible explanation for this is the self-limiting aspect
forecast for the Huntsville metropolitan area and the 5-year trans- of traffic. Using the linear extrapolation of traffic, it is quite pos-
portation improvement program 共James Moore, personal commu- sible that the trend line forecast continued, while the actual traffic
nication, April 22, 2002兲. The information was used to update the began to level off as the volume neared the capacity.
socioeconomic and roadway network. In comparison, the urban planning model has the ability to
Taking the 2001 trip generation results and the updated road- consider both socioeconomic activities and the congestion effects
way network, the remaining steps of the travel-demand forecast- while forecasting traffic into the future. The next step was to
ing process were performed in accordance with standard prac- examine the actual delay per vehicle calculated by the software
tices. The turning movement counts obtained from the urban using the different methods as illustrated in Fig. 2. Four of the six
transportation model are shown in the Table 2.
intersections resulted in less than 25% error between the urban
planning model and the actual data collected. In contrast, the
trend line analysis produced only one intersection with a percent
Results and Conclusions
difference less than 26%. Fig. 2 shows how the models are com-
The results of the case study were used to determine if the trend pared when calculating the intersection delay.
line forecasts or the urban planning model forecasts generate re- The University Drive and Jordan Lane intersection had similar
sults that are more accurate for intermediate transportation analy- results for both forecasting methods. For this intersection, the
sis. The highway capacity software 共HCS兲 package was used to trend line and urban planning model both produced a level of
perform the traffic engineering analysis for the six case study service of F, while the actual traffic counts generated a level of
intersections. The results provided by the software package in- service D. Examining the forecasted traffic for this intersection,
clude approach delays, intersection delays, approach LOS, and both methods predicted greater traffic than was actually observed
intersection LOS. The two methodologies to develop the fore- at this intersection. The additional traffic within the intersection
casted traffic counts were analyzed using HCS. Table 3 shows the might be the cause of the large difference between the delays
intersection delays and LOS for all six intersections using the two calculated using the forecasting methods and the actual counts
forecasting methodologies as well as the delay and level of ser- based on the format of the delay equation, which includes both
vice obtained using the actual counts. stopped delay and overflow delay.
From Table 3, it can be observed that the LOS for four of the After examining Table 3, qualitative analysis concludes that
six intersections in the study for the existing traffic counts and the the urban planning model forecasted traffic-turning movement
traffic counts from the urban planning model are the same. The volumes that produced results that better matched the existing