See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/282506229

Enhancement of Estimation of Internal Trip

Capture Rates for Mixed-Use Developments

Conference Paper May 2015

CITATIONS READS

0 103

2 authors:

Pei-Sung Lin Aldo Fabregas

University of South Florida Florida Institute of Technology
102 PUBLICATIONS 72 CITATIONS 30 PUBLICATIONS 22 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Pei-Sung Lin on 04 October 2015.

The user has requested enhancement of the downloaded file.

 Enhancement of Estimation of Internal Trip Capture Rates for
Mixed-Use Developments
Pei-Sung Lin1 and Aldo Fabregas2
1
Program Director, ITS, Traffic Operations and Safety, Center for Urban Transportation
Research (CUTR), University of South Florida, 4202 E. Fowler Ave., CUT100, Tampa, FL
33620; Phone: (813) 974-4910; Fax: (813) 974-5168; email: lin@cutr.usf.edu
2
Assistant Professor, Department of Engineering Systems, Florida Institute of Technology
(Florida Tech), 150 W. University Blvd., Melbourne, FL 32901; Phone: (321) 674-7184; Fax:
(321) 674-7136; email: afabregas@fit.edu

ABSTRACT
Land use of a development plays an important role in efficient city and regional transportation
planning. Mixed-use developments can considerably reduce the number of external trips than
those from a single land use development. This paper presents the results of an internal trip
capture study performed on four Mixed-Use Developments (MXDs). The study aimed at gaining
further insights on internal trip capture behavior in MXDs and how it influences traffic impacts
on public roadway systems. The results obtained confirmed that traditional methods significantly
overestimate the trips generated by MXDs. In addition, the study served as a confirmation and
enhancement of a recently proposed methodology to improve the traditional trip estimation
method for MXDs.

INTRODUCTION
Land use of a development plays an important role for efficient city and regional transportation
planning. A mixed-use development (MXD) can considerably reduce the number of external
trips than those from a single land use development. An MXD helps reduce the number of trips
on a roadway network. Internal trip capture refers to how the number of trips to and from a
development are reduced by the proximity of complementary land uses within the development
(e.g., residential to retail). Internal trips occur within the development and do not encumber the
traffic flow on the public roadways connecting the development with the transportation network.
Internal trip capture rates are used to estimate trip generation on MXDs by adjusting and
reducing single land use estimates.

Accurate internal trip capture rate estimates and methodologies are necessary to improve trip
generation estimates for MXDs. In the United States, efforts have been made to enhance both the
data availability and the estimation methodologies to estimate internal trip capture rates. In this
study, internal trip capture data from four MXDs in Florida were collected and analyzed using a
recently-enhanced internal trip capture methodology developed by the National Cooperative
Highway Research Program (NCHRP) [1]. Results showed that the enhanced methodology
produced generally more accurate estimates than those from the traditional Institute of
Transportation Engineers (ITE) internal trip capture procedure [2].

This paper first introduces the internal trip capture concept for a mixed-use development,
followed by a summary of related studies. Next, a description of the data collection methodology
and study site is provided. The results of data collection and computations of internal trip capture
rates for the four study sites using the enhanced NCHRP internal trip capture method are
presented, followed by the major research findings and recommendations. This paper, using new
data collected from this study, not only verifies the enhanced NCHRP method on accuracy of
internal trip capture estimates but also provides transportation practitioners with valuable
guidelines and clear procedures to more accurately estimate the internal trip capture rate for a
mixed-use development.

INTERNAL TRIP CAPTURE CONCEPTS

When analyzing MXDs, it is expected that the total external traffic for the MXDs will be less
than that resulting from adding together the traffic generated by the individual land uses when
considered in isolation. This trip reduction is referred to as internal trip capture. These trips occur
within the boundaries of the development using its private transportation infrastructure rather
than impacting the public roadways (see Figure 1). These internal trips are subtracted from the
aggregated trips generated by the individual developments of the MXD to obtain the external
trips. The process of estimating trip generation for an MXD can be summarized in three steps: 1)
estimate trip generation for individual land uses, 2) apply internal trip capture rate factors (trip
distribution), and 3) calculate internal/external trips (trip balancing). From this procedure, it
follows that the quality of the estimation depends on the data and methodology used to develop
accurate internal trip capture rates. The single land use estimates from ITE are corrected by
applying and internal trip capture factor. These factors are published by ITE and can be found in
Chapter 7 of the ITE Trip Generation Handbook [2]. The application of these rates is performed
via trip balancing suggested in the handbook and in NCHRP Report 684 [1].

Trip balancing is a procedure applied to land use pairs by direction such that the minimum
amount of interaction between them is determined. For trip balancing, first, the free-standing trip
generation rates are estimated by direction. For example, given land use pair AB (e.g., Office to
Retail), the outbound trips are estimated using the free-standing land use ITE rates for land use
A. Similarly, for the inbound direction (land use B), the total trips are calculated. From published
ITE internal trip capture rates, the internal trip capture percentages are selected. For Office in the
outbound direction, this is 20 percent. For the inbound direction in land use B (e.g., Retail), the
percentage of trips from Office is 8 percent. When these percentages are applied to the trip
generation estimates, the unconstrained internal trips are obtained. In the example, the origin
land use A (Office) is sending 100 trips to B (Retail); on the other hand, land use B (Retail) can
receive only 32 trips from A (Office). The minimum of these two quantities is taken as the
internal trip capture between the land use pair. This process is referred to as trip balancing (see
Figure 1). After all the internal trips are calculated and subtracted from the free-standing trip
generation estimates, the remaining quantities are the external trips or the traffic that the MXD
will generate on the public roadway system.

Roadway
External Single Land
Network
Trips Uses
B
MXD
MXD
A

Figure 1: Example of Internal Trips and Trip Balancing for MXDs.

PREVIOUS STUDIES
Accurate internal trip capture rate estimates and methodologies are necessary to improve the trip
generation estimates for MXDs. In the United States, efforts have been made to enhance both the
data availability and the estimation methodologies to estimate internal trip capture rates. In
NCHPR Report 684 [1], Bochner et al. performed a comprehensive study to enhance the internal
trip capture estimation procedure for the ITE methodology. The key contributions of NCHRP
Report 684 are the addition of AM peak hour, expansion of land uses to six, and incorporation of
proximity effects on internal trip capture. Ewing et al. [3] collected trip-making patterns from six
developments using data from the National Household Travel Survey (NHTS) and proposed a
predictive statistical model based on hierarchical non-linear models. The model used household
size, employment, gross land area of MXD, number of motorized vehicles per person in the
household, employment within MXD, and some derived indicator of the proportion of
individuals that live and work in the MXD. Lin et al. [4] conducted an internal trip capture study
in four developments in Florida following the methodology suggested by Bochner et al. in
NCHRP Report 684. The study contributed to the collective knowledge and data on trip
internalization and also served as confirmation that the proposed changes to traditional internal
trip capture methodologies could significantly improve the trip generation procedure for MXDs.
URS [5] performed an internal trip capture study in four MXDs in northeast Florida. Based on
the study, URS estimated that the internal trip capture in the developments was 20 percent in the
AM peak period and 30 percent in the PM peak period. Walters et al. [6] proposed a combination
of both NCHRP 684 [1] and EPA [7] methods to enhance the trip estimation for MXDs. The
proposed methodology, denoted by MXD+, proposed a convex combination of the outputs of the
two methods using the weights of 0.365 and 0.635 for the NCHRP 684 and EPA methods,
respectively. The combined method showed an improvement on the overall estimation error for
the study sites. A more comprehensive list of recent studies and reviews of previous trip
generation studies can be found in NCHRP 684 [1] or in Lin et al. [4].

DATA COLLECTION AND METHODOLOGY

Four MXD sites were selected to conduct detailed trip internalization study. The selected sites
varied in size and land use mix. Figure 2 presents a brief description of the sites and the area in
acres.
Creekwood (Bradenton) a suburban development with SODO (Orlando) a compact development with mid-rise
single-family detached residential units on the back end residential, medical offices, a big-box retail grocery store, and
with front-end commercial (43 acres). a variety of ground-floor retail and restaurants (18 acres).

Lakeside Village (Lakeland) a lifestyle center (open Uptown Altamonte (Altamonte Springs) combines
shopping mall) with a movie theater, hotels, and a direct existing residential, hotel, and shopping centers with new
connection to an apartment complex (74 acres). residential and a retail-themed town center (73 acres).

Figure 2: Overview of Study Sites.

Data were collected for both AM and PM peak periods and consisted of directional cordon
counts on the access points of the MXD, directional door counts, and intercept interviews.
Intercept interviews were necessary to obtain information on the internal origin-destination trip
behavior. Details of the data collection and processing can be found in Lin et al [4]. The end
result of the combination of the data for this study and previous data is a new set of internal trip
capture rates, as presented in Table 1. The left table presents the unconstrained internal trip
capture (ITC) rates. For example, from all the trips that exit an Office land use, it is expected that
20 percent will head to Retail. The table on the right presents the ITC rates for inbound trips. For
example, from all the trips entering an Office land use, it is expected that 8 percent will come
from on-site Retail. These tables are applied to the trip balancing process described in the
previous sections, selecting the minimum of these two trips for the Office to Retail trip
balancing.

Table 1: Proposed Internal Trip Capture Rates

Each previous study contributed to the basis of estimation of internal trip capture rates. For this
reason, each of these methods constitutes a testing scenario for trip generation. The list and
description of each of the trip generation scenarios or methods tested in this article are described
as follows:

ITE Single Land Use: Directional trips at the establishment level estimated using vehicle
trip generation data from Trip Generation [8]. This method assumes that all the
establishments are a free-standing single land use with no internal interaction. The MXD
traffic is calculated by adding up all the estimated trips across the establishments.
ITE Internal Trip Capture (ITC): These are the total trip generation based on the
single land use estimation with the correction procedure for internal trip capture proposed
in the Trip Generation Handbook [2]

NCHRP Report 684 (NCHRP 684): This method was proposed in NCHRP Report 684
[1] and contains data from a Florida Department of Transportation (FDOT) study in 1993
and new data collected in the NCHRP 8-51 project. A total of 93 percent of the maximum
unconstrained internal trip capture rates were from the report and about 7 percent from
the 1993 FDOT study.

NCHRP 684 with Proximity: This method is similar to the NCHRP 684 method
described previously with the addition of a proximity adjustment factor to improve the
prediction capability of the estimator for large MXDs.
 NCHRP+FDOT 2014: This method contains data from the studies in NCHRP Report
684 [1] including the 1993 FDOT study and adds data from the described study. In total,
sources of the maximum unconstrained internal capture rates are 64 percent from
NCHRP data, 33 percent from FDOT 2014 data, and 3 percent from FDOT 1993 data.

NCHRP+FDOT 2014 with Proximity: This method is similar to the NCHRP+FDOT

2014 method described previously with the addition of proximity adjustment factors from
NCHRP Report 684 to improve the prediction capability of the estimator.

The PM peak period (4:006:00 PM) was selected as the period of interest for analysis purposes
due to more meaningful activity levels. The selected dependent variable for the trip generation
method was the hourly two-way external vehicular trip generation for the MXD. The collected
data were used to developed origin-destination matrices for all the study MXDs. These rates
were used to predict the observed cordon counts for the study sites. Deviations with respect to
the observed cordon counts were used as the performance measure of the estimation procedure
for the individual MXDs. Figure 3 presents the results of the analysis to determine the
effectiveness of the proposed methods to estimate trip generation of MXDs. The dotted line
represents the observed bidirectional cordon counts or 100 percent. The prediction errors are
expressed as percentages of the observed cordon counts. Values greater than 100 percent
represent overestimation, and values below 100 percent represent underestimation. The
quantitative results and obtained and the best estimates are presented in Table 2.

240%
Percent of Observed Cordon Count

220%
200%
180%
160%
140%
120%
100%
80%
60%
40%
20%
0%
Atlantic Boca Del Country Creekwood Lakeside Legacy Mockingbird SODO Village
Station Mar Isles Town Station Commons
Center
ITE Single Land Use ITE ITC
NCHRP NCHRP with Proximity
NCHRP+FDOT 2014 NCHRP+FDOT 2014 with Proximity

Figure 3: Comparison of Cordon Count Estimates (Bidirectional)

Using Combinations of Internal Trip Capture Studies.
 Table 2: Results for Best Estimates during PM Peak Period
Development NCHRP (684) NCHRP+FDOT 2014 Area Best Estimate
Atlantic Station 100% 103% 117 NCHRP
Boca Del Mar 94% 103% 296 NCHRP+FDOT 2014 with Proximity
Country Isles 92% 96% 71 NCHRP+FDOT 2014 with Proximity
Creekwood 101% 99% 43 NCHRP or NCHRP + FDOT 2014 (tie)
Lakeside Village 92% 96% 74 NCHRP+FDOT 2014 with Proximity
Legacy Town Center 101% 89% 77 NCHRP with Proximity
Mockingbird Station 116% 128% 11 NCHRP
SODO 165% 164% 18 NCHRP+FDOT 2014
Village Commons 112% 111% 101 NCHRP+FDOT 2014

The ITE single land use estimates are presented for reference. These estimates contain all the
potential trips that a single land use can generate and are the basis for trip generation for MXDs.
It is observed that the single land use trip rates severely overestimate the trip generation in the
MXD. The ITE ITC method is a profession-accepted traditional internal trip capture method
based on data from the FDOT 1993 studies. The ITE ITC adjusts the single land use estimate to
reflect trip reductions due to internal trips and consistently overestimates the trip generation for
MXDs. The method proposed in NCHRP Report 684 [1] provides a more consistent estimate for
MXD trip generation estimates. NCHRP with proximity included a correction for proximity that
helps to account for internal trip capture decay in large developments. The combined data
approach (NCHRP+FDOT 2014) using the revised maximum unconstrained internal trip capture
rates improved the prediction capability of the existing data-method combination in five out of
eight test cases, with one test case tied. The data of the four sites were joined with NCHRP 684
data. In the resulting dataset, the newly added data contributed to 30 percent of the highest
inbound rates and 33 percent of the outbound rates. While this seems to be a slight improvement,
without additional data it is challenging to determine whether this variation is due to randomness
or to true cause-effect relationships. On the other hand, this small variation could be an
indication of convergence, and the obtained rates hover around the true rates. This will make the
know dataset generalizable or applicable to MXDs of similar characteristics such as size between
18 and 300 acres. In terms of proximity adjustments, notice in Table 2 that for Mockingbird
Station, SODO, and Creekwood, the best estimate was not adjusted for proximity. These
developments have an area under 50 acres. This is an encouraging result that may lead to a new
criteria or refinement for existing trip generation methods for MXD. This can be further explored
should similar studies become available.

CONCLUSIONS AND FUTURE WORK

This study focused on data collection and verification of a series of methodologies to estimate
trip generation on MXDs. The results obtained confirmed that traditional methods significantly
overestimate the trips generated by MXDs. The proposed improvements to traditional methods
such as those in ITE Trip Generation Handbook provided more realistic estimates. The sites
surveyed in this study contributed to the knowledge base on internal trip capture data. The new
 data contributes to 30 percent for outbound rates and 33 percent for inbound rates when the
highest interaction criterion is used to select a single set of unconstrained internal trip capture
rates. The combined data approach (NCHRP+FDOT 2014) using the revised maximum
unconstrained internal trip capture rates improved the prediction capability of the existing data-
method combination in five of eight test cases, with one test case tied. For MXDs with areas of
43 acres or more, the estimation errors ranged from 89112 percent. Two cautionary notes are
worth mentioning. The first is that updated methods may underestimate the actual traffic. More
fine-tuning of the existing methodologies is required such that more conservative estimates are
more likely to occur. The second relates to the application of internal trip rates to highly compact
MXDs (under 20 acres) such as SODO or Mockingbird Station. The data and method applied
suggest that in those cases there is a significant overestimation of the actual trips. More studies
on highly compact development may be required to properly address this issue. Data suggested
that proximity adjustments may be beneficial for developments over 50 acres. This is an
experimental encouraging result that may lead to refinements for existing trip generation
methods for MXDs. A repository of validation data for MXDs should be developed for use in
evaluating the predictive capability of current internal trip capture methods. Data should consist
of land use inventory, cordon counts, and door counts, at minimum. Additional origin-destination
interviews are highly desirable to obtain further insights. Technology can also help in
establishing internal trip patterns using mobile devices and social media network analyses.

REFERENCES
[1] B. Bochner, K. Hooper, B. Sperry, and R. Dunphy, NCHRP Report 684, "Enhancing Internal
Trip Capture Estimation for Mixed-Use Developments," National Cooperative Highway
Research Program, Washignton, DC, 2011.
[2] Institute of Transportation Engineers, Trip Generation Handbook, 2nd Edition: An ITE
Recommended Practice, Washington, DC: Institute of Transportation Engineers, 2004.
[3] R. Ewing, M. Greenwald, M. Zhang, J. Walters, M. Feldman, R. Cervero, L. Frank, and J.
Thomas, "Traffic Generated by Mixed-Use Developments- Six-Region Study Using
Consistent Built Envirionmental Measures," Journal of Urban Planning and Development,
137(3), pp. 248-261, 2011.
[4] P. Lin, A. Fabregas, A. Pinjary, K. Seggerman, C. Lee, V. Koneru, B. Bochner, and B.
Sperry, "Trip Internalization in Multi-Use Developments," Report No BDK84-977-10,
prepared for the Florida Department of Transportation, 2014.
[5] URS, "Internal Trip Capture Study District 2." Report prepared for the Florida Department of
Transportation," 2010.
[6] J. Walters, B. Bochner, and R. Ewing, "Getting Trip Generation Right: Eliminating the Bias
Against Mixed Use Development," American Planning Association, 2013.
[7] Environmental Protection Agency, "Trip Generation Tool for Mixed-Use Developments,"
[Online]. [Accessed 2013].
[8] Trip Generation, 8th Edition: An ITE Informational Report, Washington DC: Institute of
Transportation Engineers, 2008.

View publication stats