Measuring the Air Quality and Transportation
Impacts of Infill Development
ir.
Ti
as-
7
Pn>l»elion Agency
-------
ni
United States
Environmental Protection Agency
(1807-T)
Washington, DC 20460
Official Business
Penalty for Private Use $300
EPA231-R-07-001
November 2007
www.epa.gov/smartgrowth
:. • •
-------
Acknowledgements
The U.S. Environmental Protection Agency (EPA) initiated this project to quantify the air quality emissions
impacts from regional brownfield and infill development policies. Industrial Economics, Inc., in association with
Cambridge Systematics, Inc., was selected by EPA to perform this work in collaboration with the Metropolitan
Area Planning Council (Boston), Denver Regional Council of Governments, and the City of Charlotte. The report
was reviewed by a panel of transportation planners: Ken Cervenka, North Central Texas Council of
Governments; Gordon Garry, Sacramento Council of Governments; and Scott Lane, The Louis Berger Group.
To request additional copies of this report, contact EPA's National Center for Environmental Publications at 800-
490-9198 or by email at nscep@bps-lmit.com and ask for publication number EPA 231-R-07-001.
Front Cover Photos:
Jefferson North End, Dallas, Texas. This 540 unit residential complex, opened in 1998, was built on a site that had
been vacant for 20 years. (Photo Courtesy of US EPA, Office of Brownfields and Land Revitalization)
Back Cover Photos:
Southside Neighborhood, Greensboro, North Carolina. This mixed use redevelopment is within a five to ten
minute walk from the central business district. It includes 30 single-family homes, 10 two-family homes, 50
townhouses, 10 restored historic homes, and 20 live/work units. (Photo Courtesy of City of Greensboro,
Department of Housing and Community Development)
-------
Measuring the Air Quality and Transportation Impacts of
Infill Development
-------
Executive Summary
Infill Development as a Key Transportation Management Strategy
Many regions are struggling to balance transportation needs with community
revitalization and environmental protection. The potential for infill development to
support all three goals is what sets it apart as a unique strategy. While the positive impact
of redevelopment projects may be readily apparent at the community level, their regional
transportation and air quality benefits can be harder to quantify.
Fundamentally, well designed neighborhoods in more accessible places make walking,
biking and transit more convenient options. Therefore, policies that increase the amount
of urban and suburban infill development can help more people meet their everyday
needs with less driving. In turn, this can reduce traffic and contribute to better regional
air quality.
The complicated nature of redevelopment often requires the public sector to act as a
catalyst by providing financial subsidies, assembling land or upgrading infrastructure.
The direct economic benefits justify many public investments, but more substantial
commitments could be supported if the indirect transportation and air quality benefits
could also be quantified.
This study illustrates how regions can calculate these benefits. The basic approach relies
upon standard transportation forecasting models currently used by Metropolitan
Planning Organizations across the country. The results suggest that strong support for
infill development can be one of the most effective transportation and emission reduction
investments regions can pursue.
The Report's Purpose: Demonstration of Methods
Although less vehicle travel and fewer emissions are reasonable outcomes to expect from
infill development, quantifying such benefits has proved challenging. In most cases, the
forecasting models used for regional transportation planning are not set up to capture the
effect of innovative land use strategies. Therefore, they typically do not capture the
changes in vehicle travel generated by increasing development in walkable communities
with convenient access to transit. Quantifying benefits is also complicated by the need to
establish baseline development trends. In other words, measuring the net benefit of a set
of infill projects requires establishing where development would have otherwise gone.
This report summarizes three case studies, each testing slightly different approaches to
these analytical problems within traditional four-step travel-demand models. The
analysis shows how standard forecasting tools can be modified to capture at least some of
the transportation and air quality benefits of brownfield and infill development.
11
-------
Some of the key modifications include:
• Indices to reflect how mixed use development changes travel patterns
• Adjustments to account for shifts to non-motorized travel
• Smaller analysis zones in the models to capture the impact of neighborhood land
use patterns
• Emissions estimates based on the number of car trips as well as by distance
traveled
Key Findings
Across the three case studies, redirecting jobs and households to brownfield and other
infill sites reduces overall travel, congestion and emissions from cars. For example, if just
8 percent of Denver's jobs and households were shifted over time toward 10 regional
centers, congestion would be reduced by over 6 percent and emissions would be reduced
by about 4 percent. This would be equivalent to removing nearly half a million trips per
day from the region's roads, a significant share of the daily average (12.7 million miles). If
the same amount of development was concentrated in 31 locations, the reduction in
emissions would be somewhat smaller (3 percent).
The Charlotte case study evaluated the impact of increased infill development in a single
corridor. Although, a much smaller number of jobs and homes were relocated to infill
sites, the analysis demonstrates the benefits of focused development around transit.
While the new rail service alone did reduce congestion in the corridor, it had a minimal
impact on the region's emissions. However, when 16,000 households and 10,000 jobs are
relocated near the South Corridor stations, the reduction in emissions was 10 times greater
and transit ridership increased by more than 6,000 trips each day.
In Boston, the analysis considered redevelopment in just 13 suburban towns along the I-
495 Corridor. Redirecting new development to brownfield sites in these towns reduced
vehicle travel by 154,000 miles during the evening rush hour. Given the corridor's
average car trip of 15 miles, this reduction is equivalent to eliminating more than 10,000
trips. In road capacity terms, this close to the additional trips accommodated by specific
road-widening projects proposed for these towns. These projects are expected to cost $5
to $17 million each. (Boston Metropolitan Planning Organization 2006)
Compared with other policies adopted to meet regional air quality goals, these reductions
are both significant and cost effective. For example, the reductions generated by projects
funded under the Congestion Mitigation and Air Quality (CMAQ) program typically cost
around $20,000 per year for each ton of pollution they eliminate. (FHWA 2006, Appendix
4) With such a benchmark, Denver could spend roughly $17 million to facilitate the kind
of infill development evaluated in this study and still be more cost effective.1
1 Based on the assumption that land use related reductions are maintained over a 30-year period,
the rough lifetime of many commercial properties. Residential development tends to have a much
longer lifespan and would likely produce VMT reduction benefits over a much longer period.
However, a more conservative assumption was made for a basic cost-effectiveness illustration.
m
-------
Intended Audience and Structure of the Report
The report is intended for both land use and transportation planners, to act as a bridge
between innovative redevelopment policy and transportation modeling. The first part of
the report outlines the issues, methods, and findings of each case study in general terms.
The goal is to help non-technical readers understand how the benefits of infill
development might be measured using traditional transportation planning tools.
Technical appendices cover the details of each regional analysis at the level of detail
needed to design and launch a comparable study. However, as noted by the peer
reviewers, the complexity of travel demand models and their unique evolution in each
region mean that even the technical appendices are not exact instructions for replicating
the analysis. Rather, they illustrate methodologies that must be customized to fit each
region. On the other hand, they do describe specific concepts and techniques that can
serve as a starting point for modifying most regional travel demand models.
IV
-------
Table of Contents
Executive Summary ii
1.0 Introduction 1
1.1 Brief Overview of the Case Studies 2
1.1.1 Boston 2
1.1.2 Denver 5
1.1.3 Charlotte 7
1.2 Key Findings 11
2.0 Study Design: Lessons Learned 13
2.1 Limitations of Traditional Travel Demand Models 13
2.1.1 Aggregate nature of models 14
2.1.2 Non-motorized travel 14
2.1.3 Trip chaining and tour based modeling 14
2.2 Emissions Analysis Issues 15
2.3 Evaluation Measure s 15
2.3.1 Measures of the Amount of Travel 15
2.3.2 Measures of Transportation Level of Service 16
2.3.3 Measures of Environmental Impacts 17
3.0 Comparison of Results 17
3.1 Land Use and Vehicle Miles Traveled 17
3.2 Congestion and Speed 18
3.3 Emissions 18
3.4 Effects of Increased Development 19
3.5 Conclusions and Implications 22
3.5.1 Reductions in Vehicle Travel and Emissions 22
3.5.2 Use and Limitations of Travel Demand Models 22
3.5.3 Relative Contribution of Land Use Strategies 23
3.5.4 Future Research Needs 24
Detailed Technical Appendices
Appendix A - Boston
Background A-l
Detailed Results A-2
Transportation Analysis Methods A-4
Intrazonal VMT Adjustment A-5
Emissions Analysis Methods A-7
Appendix B - Charlotte
Background B-l
Detailed Results B-2
Transportation Analysis Methods B-6
Emissions Analysis Methods B-9
v
-------
Appendix C - Denver
Background C-l
Detailed Results C-5
Transportation Analysis Methods C-6
Intrazonal Trip Changes C-7
Development and Calculation of Mixed Land Use Indices C-7
Calculation of Zonal MUIs for Year 2001 and Year 2025 C-l 1
Calculation of Intrazonal Trips C-l 5
Intrazonal Trip Adjustment Models C-l5
Revision of Mixed Land Use Indices C-20
Model Results Discussion and Recommendations C-20
Trip Generation Revision C-20
Pedestrian or Bicycle Friendly Zones in Boulder County C-20
Trip Adjustment Factor Calculation C-21
Emissions Analysis Methods C-22
Appendix D - Benefits of Using MOBILE6 D-l
VI
-------
1.0 Introduction
Local governments across the country are paying more attention to infill development. Former
industrial sites, declining suburban malls, vacant properties and other underutilized land all
provide opportunities for redevelopment. Projects developed on_such sites are often pursued
for their economic development benefits. However, redeveloping underutilized land in cities
and suburbs also has the potential to reduce vehicle travel and contribute to better air quality.
In fact, the studies summarized in this report suggest that actively supporting infill
development can be a highly effective regional transportation policy. If done well,
redevelopment creates neighborhoods where residents can accomplish their daily activities
with less driving. Previous site level studies suggest that shifting development to more
accessible locations reduces vehicle travel per person by 30 to 60%2.
At a regional level, a significant number of well designed infill projects can go a long way
toward helping meet air quality goals. Specifically, such changes can make important
contributions to the emission reduction targets contained in State Implementation Plans. These
SIP plans are critical because they determine if future transportation investments conflict with
regional air quality goals defined under the Clean Air Act.
This study quantifies the air quality benefits of regional growth scenarios that increase
development on brownfield and other infill sites. To achieve this objective, three Metropolitan
Planning Organizations (MPOs) incorporated new analytical components into their existing
travel demand forecasting tools. Their models were enhanced to reflect techniques described in
EPA's Comparing Methodologies to Assess Transportation and Air Quality Impacts of Brownfield and
Infill Development report. Various publications from the U.S. Department of Transportation's
Travel Model Improvement Program (TMIP) also served as a key resource in the process. The
goal was to develop techniques for estimating emission reductions that would be both
transferable and acceptable within a State Implementation Plan (SIP), conformity determination,
or ozone flex plan.
However, achieving this objective required overcoming a few major obstacles. First and
foremost, the analytical framework3 at the heart of nearly all regional transportation planning
models has difficulty capturing interactions between land use and transportation systems.
Even when MPOs have incorporated land use feedbacks into their travel demand models, they
tend to be regional in nature and fail to capture the key neighborhood level characteristics. It is
these smaller scale land use patterns that often contribute most to the reduced driving expected
from well designed infill projects. Other common limitations include: only examining work-
related travel, not considering walking as a mode of travel, and including very little detail on
land use characteristics between "travel analysis zones."
2 Ewing, R. and R. Cervero. 2001. Travel and the Built Environment: A Synthesis. Transportation
Research Record 1780, 87-114.
U.S. EPA, 1999, The Transportation and Environmental Impacts of Infill Versus Greenfield
Development: A Comparative Case Study Analysis, EPA publication number 231-R-99-005.
3 Nearly all regional travel demand models follow the "four step" process. Typically this is a set of
connected models that estimate the following- (1) trip generation, (2) trip distribution, (3) mode
choice and (4) traffic assignment.
-------
Staff from the three participating transportation agencies performed the comprehensive
analyses. They included the metropolitan planning organizations in Boston and Denver and the
City of Charlotte. Each worked with EPA to test whether a specific set of infill projects in the
region could reduce emissions relative to current development trends. They used their existing
travel demand models, along with EPA's MOBILE6 emissions software, to measure the benefits
of compact infill development scenarios. The analysis also produced transportation outcome
measures such as person trips, average trip lengths, vehicle miles traveled, vehicle hours
traveled, transit mode share, average speed, and congestion. In each case, the analysis found
that increased brownfield and infill development would result in substantially lower emissions
of hydrocarbons (VOCs), nitrogen oxides (NOx), and carbon monoxide (CO).
The case studies examined projects at differing regional scales. In Boston, infill development
was examined in a small portion of the metropolitan area. It tested how changing growth
patterns in one corridor could improve the air quality outlook of the region as a whole. Denver
examined the entire region to see how focusing development in a few large urban and suburban
centers would compare to current development trends. Charlotte focused on the impact of infill
development concentrated around a new light-rail transit line.
1.1 Brief Overview of the Case Studies
1.1.1 Boston
This regional case study evaluated the impact of reorienting development patterns in 13 towns
along the 1-495 corridor, roughly 20 miles west of downtown Boston. The Metropolitan Area
Planning Council (MAPC) and the Central Transportation Planning Staff (CTPS) ran two
scenarios:
1) Current development trends continued
2) Concentrated development in town centers and interchanges along 1-495.
Both scenarios assumed the same amount of employment and household growth. However,
under the Concentrated Growth scenario, about 18,000 households (14 percent of the corridor's
total) and 102,000 jobs (22 percent of the total) would be redirected over time into more
accessible locations. The specific parcels to be redeveloped in the scenario were Ashland Center
and Hopkinton Center, two large brownfield sites, and a number of smaller infill sites near the
town centers and freeway interchanges in the other 11 towns.
The MAPCs travel demand model found that more accessible development would produce
significant transportation benefits. The analysis forecast less driving and traffic congestion as
well as more trips made by transit or walking. Overall, people who live and work in the
corridor would drive 239,000 fewer miles each day during the evening rush hour. These
changes would result in significant air quality benefits for each town and the region as a whole.
Compared to current development patterns, shifting growth to the more accessible sites would
result in 5 to 8 percent lower emissions of VOVs, NOx and CO by (see Figure 1.1).
The significance of these reductions is best understood when compared to typical transportation
demand management and emission reduction policies. Effective regional strategies are
generally comprised of many actions that together add up to a significant reduction of regional
-------
emissions. It is unusual for any single action to reduce regional emission by more than 1
percent, but 10 to 20 actions together can achieve the air quality targets established under the
Clean Air Act. Similarly, individual demand management policies cannot solve traffic
problems, but a comprehensive set of policies do have an impact. The reductions in driving and
emissions that would result from putting more homes and businesses in the corridor's town
centers exceeds many of the demand management actions outlined in the MAPC's Regional
Transportation Plan. This is particularly significant in light of the more than 80 jurisdictions for
which no land use changes were considered.
Figure 1.1 -Reductions from Focused Redevelopment in the 1-495 Corridor
Shift of Development
toward Infill Areas...
Results in the following changes in
travel and emissions
25% n
D Share of 13 Town Total
D Share of Regional Total
-10%
Households Jobs
Vehicle Congestion
Miles
VOC
NOx
-------
Figure 1.2 Towns in the Boston Case Study
TOWN SEND PEPPERELL
AI.EM & MJQKUHIAJ
ACTON / / BEDFORD
CONCOR
I49'
V
BER11N* HUDSON J SUDBUBY
BOROUGH ISQym-. fSAMNGHAM
ERBURN / DOVER
EAST
EASTON \ TOST I BRIDGE-
-------
Specific modifications to the region's travel demand model were required to examine the
potential benefits. First, the model's traffic analysis zones were redrawn to significantly
increase the level of detail (from 54 zones to 137 zones). This change was critical to capturing
the effect of moving development to town center locations. The previous zones often grouped
these suburban centers and their surrounding town areas into the same zone. The staff also
developed a mixed-use index indicator to better capture the travel demand impact of co-locating
houses, jobs, shopping, and entertainment. This variable was intended to reflect how modest
changes in the convenience of daily activities can substantially increase the share of trips made
on foot or transit rather than by car. However, the changes in vehicle travel primarily reflect a
shift toward shorter car trips rather than substantial increases in walking or transit use. This is
not surprising given that the infill projects in the corridor were still surrounded by
predominantly low density communities with limited transit services.
1.1.2 Denver
Analysts at the Denver Regional Council of Governments (DRCOG) examined three land use
scenarios. The scenarios evaluated differences in driving, congestion, and emissions over a 30
year timeframe. The first extended current development trends, with a majority of new homes
and employment widely dispersed across suburban locations. The second focused
development into 31 mixed-use centers. The third focused development into 10 higher-density
mixed-use centers (Figure 1.3). It is important to note that both alternative development
scenarios assumed substantial infill in suburban locations. Together these two alternative
scenarios would shift about 7 percent of all households and 9 percent of all jobs into more
compact areas.
Figure 1.3
Forecasted Employment
10 Center Scenario
Total Employment
Less Than 500
501 - 1,500
1,501 -3,500
3,501 - 5,000
Over 5,000
Boulder
Broomfield
CBD
5 Cherry Creek
6 Englewood Town Center
Gates Redevelopment
8 Lakewood
9 Longmont
10 Stapleton Redevelopment
-------
When the travel demand model was run, they showed focusing development could
substantially reduce travel at a regional scale. The 31-Center approach could reduce driving by
more than 2.7 million miles per day, about 3 percent of the regional total. Focusing new growth
into a smaller number of high-density centers - the 10-Center scenario - would reduce the daily
vehicle travel in the region by 3.6 percent. In both cases, more focused development lowered
congestion and raised average travel speeds compared to the trend scenario. Both scenarios
would also increase the share of transit trips - by more than 11 percent in the 10-Center scenario.
The changes in vehicle travel from the scenarios were also applied to a regional emissions
model that estimated reductions in VOC, NOx, and CO emissions by 3 percent to 4 percent
(Figure 1.4).
Figure 1.4 -Reductions from Focused Redevelopment in Regional Mixed Use Centers
Shift of Development
toward Infill Areas...
Results in the following changes in
travel and emissions
Households
Jobs
Vehicle Congestion VOC
Miles
NOx
-------
The Denver analysis also relied upon modifications to the existing regional travel demand
model. First, better estimation of non-work trips was achieved by more accurately
representing the attractors of such travel, such as retail centers and major institutions. Second,
the original model limited mode choice to car or transit and did not incorporate walking to bike
trips. Since a greater share of trips made on foot or by bike is a primary way that infill
development reduces driving, this modification of the model was critical. Adjustments to the
mode choice component of the travel demand model were tested in a portion of the DRCOG
modeling area - Boulder County. Staff divided the county into three zones - from least to most
friendly for bicycling or walking. Travel records for these areas were examined in detail. In
Zone 1, people walked or bicycled for 40 percent of commuting trips and about 44 percent of
errands run from home. At the other end of the spectrum, in Zone 3, just 4 percent of
commuting trips and 10 percent of errand trips were via foot or bicycle (see Figure 1.5). Based
on the relationship between key land use characteristics in these three zones and rates of
walking and biking, the team calculated adjustment factors for the rest of the region's zones.
Figure 1.5 - Impact of Land Use on Non-Motorized Travel in Boulder, CO
Impacts of Land Use on
Non-Motorized Travel in Boulder, CO
zone 3
zone 2
zone 1
0%
10%
20%
30%
40%
50%
Percent trips via bike, walk
1.1.3 Charlotte
A third case study was conducted by the city of Charlotte, which also maintains a regional
travel demand model. While Denver and Boston each assumed the transportation
infrastructure would essentially stay the same, Charlotte wanted to examine the transportation
impacts of infill development in relation to their proposed light-rail system. The city conducted
its analysis while the South Corridor was still in the planning stages. The project has since
moved forward and the rail line is scheduled to open in the Fall of 2007.
-------
Figure 1.6 - Charlotte Transit Corridors with Study Area
2025 Household Density
Map Layers
J.'.l Counties
..... CorrtBor
Transit Corridors
2025 HH / Sq.M
0 to 499
53CO •: 200DO
Other
Highways
— Ttiorci-ghfare
Other
1 i
-------
Charlotte looked at four scenarios. The Baseline assumed no rail line and a continuation of
current development trends. The Transit scenario assumed the light-rail line would be built, but
with no change in current development patterns. The Transit-Station Infill scenario projected
building light rail and concentrating more development around the 12 station areas in the South
Corridor. Finally, the Transit-Station Infill-Road Improvement scenario also added improvements
to the street network around the station areas. The two infill scenarios assumed that about 4
percent of households in the region would eventually be shifted into the South Corridor,
tripling the number of households in the areas surrounding the new light-rail stations. Under
these scenarios, a much smaller share of regional employment would be redirected to station
areas, but amount of business activity near the stations would still increase by about 50 percent.
Evaluating four scenarios provided an opportunity to isolate the impact of each on vehicle
travel, congestion and emissions in the region. The analysis separately measured the impact of:
1) simply adding transit, 2) adding transit and changing land use patterns, and 3) making street
improvements near the station areas to accommodate the new development. The model also
measured the impact at three different geographic scales: the immediate vicinity of the stations,
the entire South Corridor, and throughout Mecklenburg County.
Organizing the scenarios in this manner highlighted important results. For example, adding the
rail line without making land use changes would increase transit use in the corridor (1,000 trips
per day), but only have modest impact on vehicle travel at the county level. However, when
combined with infill development, transit ridership jumped by more than 7,000 trips per day
and reduced overall travel in the county by 2 percent (Figure 1.7). On the other hand, moving a
significant number of homes and jobs to the station areas did increase vehicle travel and
congestion at the neighborhood level (10 to 15 percent). Therefore, the fourth scenario
examined whether making road improvements in the station areas mitigated these impacts. In
short, the road improvements further increased vehicle travel, but mitigated the localized
congestion that accompanied the infill development. This produced slightly lower emissions
relative to station area development without local road improvements.
The changes in emissions followed a similar pattern. The biggest emission reductions came
from supporting infill development around the rail stations. While the two infill scenarios
slightly increased emissions within the corridor, they significantly reduced emissions at the
regional level (See Figure 1.8). Since the precursor emissions responsible for ozone pollution
(VOC and NOx) are generally not harmful until they combine in the atmosphere, the county
emissions total is outcome measure with the most direct implications for public health. Further,
making road improvements along with the station area development substantially offsets the
local increase in congestion and emissions.
-------
Figure 1.7 Emission Reductions in Charlotte
Projected Emission and Travel Reductions from
Charlotte South Corridor Alternatives
0.0%
-2.5% J
I Trasnit Only
I Station Area Redevelopment
I Station Redevelopment with
Road Improvements
VMT (OOO's)
VOC (kg)
CO (kg)
NOx (kg)
Figure 1.8 - More Travel and Emissions in Corridor Offset by Regional Benefits
Shift of Development
toward Infill Areas...
Results in the following changes in
travel and emissions
3.0%
2.0%
S
£ 1.0%
re
M 0.0%
4)
{£.
- -1.0%
re
-2.0%
g -3.0%
<5
o_
-4.0%
-5.0%
Households Jobs
D Mecklenberg Coounty
• South Corridor
Vehicle Congestion
Miles
VOC
CO
NOx
10
-------
Previous Studies and Guidance - The Atlantic Station Project (Atlanta, GA)
EPA first examined the potential air quality benefits of infill development for the
proposed 135-acre Atlantic Station project in downtown Atlanta. A bridge was
needed to connect the development with a nearby MARTA rail station. However,
due to air quality constraints, the regional transportation agency could not add
the bridge to its capital construction plan. Before the project could move forward,
the transportation agency had to demonstrate the net gains in air quality it would
produce.
EPA's analysis compared the emissions impacts of the new development to the
same number of homes and jobs located on typical sites in suburban jurisdictions.
The analysis found that developing the abandoned Atlantic Steel site would
substantially reduce driving relative to alternative sites on previously
undeveloped land. Shorter car trips alone were estimated to reduce overall
vehicle travel by 14 to 50 percent. Additionally, if the new community were
designed to encourage bicycling, walking, and transit, driving would be reduced
another 5 percent. These changes were then coupled with a regional emissions
model to quantify the reductions in key pollutants. As a result of this analysis,
the development was designated a Transportation Control Measure in the
region's air quality plan. EPA has since issued the following guidance on
methodologies communities can use to document the air quality benefits of infill
development:
Comparing Methodologies to Assess Transportation and Air Quality Impacts of
Brownfield and Infill Development, EP A-231 -R-01 -001
Granting Air Quality Credit for Land Use Measures:
Policy Options, EPA SR99-09-01
1.2 Key Findings
The three case studies demonstrate - across a range of scenarios and regional contexts - that
redirecting development to more walkable, transit accessible areas reduces driving and
emissions. Shifting 5 to 10 percent of a region's homes and jobs to infill locations was estimated
to produce 2 to 5 percent less vehicle travel and a 3 to 8 percent reduction in emissions (Figure
1.9).
The majority of the reductions were due to shorter vehicle trips. While this is a key finding
consistent with previous studies, it also reflects a weakness shared by the three analyses. In
spite of modifications, the travel demand models still did not have the ability to examine how
good site location and design might increase rates of walking and biking. For example, the
Charlotte model was only able to consider personal vehicles and transit as travel modes.
Although a mixed-use and mode choice analysis was conducted in Boston, it was not included
in the final results. These are important analytical shortcomings since research4 shows that
community design significantly influences the amount of bicycling and walking. Therefore, the
f Land Use and Site Design: Traveler Response to Transportation System Changes, TCRP Report 95 Chapter 15
11
-------
VMT reductions and emissions benefits documented in this project are most likely
underestimated.
On the other hand, the project did demonstrate how regions can use existing tools to measure
the relationship between redevelopment and air quality. The travel demand models remain
limited in their ability to account for small scale changes to the built environment. However,
modifications to the models can help mitigate some of these problems. Communities interested
in using existing models should consider:
• Smaller zone sizes for analysis in order to capture intrazonal trips at a finer land
use scale
• Adjusting mode choice models to capture for increased walking and cycling
• Using indices to represent the degree of mixed use and its impact on travel patterns
• Estimating emissions by number of trips as well as by distance traveled
This report shows that directing new growth into reclaimed brownfield and infill sites can help
meet their need for growth while addressing regional air quality issues. While still limited,
existing transportation planning tools can help quantify the potential impact of these changes.
This is an important step in allowing decision makers to choose the best course for the future of
their region.
Shift of Development
toward Infill Areas-
Results in the following changes in
travel and emissions
10% n
D Denver (10 Regional Centers)
• Charlotte (Single Rail Corridor)
D Boston (13 Suburban Towns)
Households
Jobs
Vehicle Congestion VOC
Miles
NOx
Figure 1.9 - The Effect on Travel and Emissions from Shifting Development to Infill Sites
12
-------
2.0 Study Design: Lessons Learned
While the MPOs in this study had a variety of objectives, they shared a common purpose - to
better capture the benefits of brownfield and infill development. Each had sketched out plans
for quantifying the impacts of alternative land use patterns. However, they differed in the scale
and context for their analysis.
• Boston - Examined a small part of the metropolitan area to test the impact of local
redevelopment on the region as a whole
• Denver - Modeled region-wide shifts in development to see how redirecting
growth toward key regional mixed use centers compared with the current pattern
of unfocused growth
• Charlotte - Examined the impact of infill development around the station areas of a
proposed light rail corridor
They also made different assumptions about their regional transportation systems. Boston and
Denver assumed no changes to the system, even though shifting land use patterns often create
pressure for road capacity expansion. Based on these simplified assumptions, neither study
was able to estimate the secondary effect of changes in transportation supply associated with
alternative development patterns. In contrast, the Charlotte analyses examined the effects of
changes in both the transit and street networks associated with the proposed land use scenarios.
2.1 Limitations of Traditional Travel Demand Models
All three of the partner communities use traditional four-step models, but have incorporated
some features not typically found in other regions. For example, each considers the feedbacks
between the transportation network and regional land use patterns. They also take some steps
to consider the impact of land use patterns on travel behavior. Even with such modifications,
the four-step approach still has some important limitations when evaluating the travel demand
impacts of infill development.
The follow sections discuss these specific limitations. However, before discussing the critiques
in detail, it is important to first describe the basic components that make up a traditional travel
demand forecasting models:
1. Trip generation - The number of person trips by purpose generated in each traffic
analysis zone. Estimated based on the amount of activity in each zone, defined by
population, households, and employment.
2. Trip distribution - The trips generated from each zone in step 1 are distributed to
create person trips from origins to destinations. They are often segmented by travel
purpose (e.g. work vs. non-work trips).
3. Mode choice - The trip tables estimated in step 2 are split into different travel modes
based on the characteristics of the trip (purpose, distance, etc.).
4. Trip assignment - The vehicle and transit trips resulting from the mode choice model
are loaded onto the highway and transit networks to produce volumes on roadways
and transit ridership estimates. Walking and bicycle trips are seldom modeled
directly.
13
-------
2.1.1 Aggregate nature of models
Because the four-step process was conceived at a time when computers were much less
powerful, it was necessary to simplify the way in which urban areas were modeled. One
simplification was the aggregation of regions into traffic analysis zones (TAZ). With a few
exceptions, all estimated travel is based on the average characteristics for each TAZ. Trips may
begin or end at different points within a zone, but are tied to the same average land use and
household characteristics. This introduces an aggregation error into the model. Over the years,
as computing capabilities increased, models were able to have a much greater number of zones,
diminishing—although not eliminating—the effect of aggregation error. Currently, a typical
modeling region may consist of several hundred zones or several thousand zones in a larger
metro area.
The treatment of trips that stay within a single zone is one key issue where using TAZs as the
basis for analysis becomes a critical limitation. Since intra-zonal trips cannot be assigned to a
traditional model's transportation network, the system is calibrated so travel within a TAZ
matches observations in the baseline year. In scenarios where land use patterns become
significantly more concentrated, this calibration will probably underestimate intrazonal travel.
The necessary aggregation of trip end locations into zones also has implications for the type of
analysis required for this project. Any concentrated development scenario must be represented
by the same zone system as in the baseline scenario. Therefore, if new development were
concentrated in zone near a transit station, the model results would still be based on the
assumption that growth was spread evenly throughout the zone. However, in the Boston
analysis, MAPC and CTPS recognized the constraints of the zone system used in their model
and developed a more fine grain set of zones for the 13 towns. This reduced the aggregation
error associated with the analysis.
2.1.2 Non-motorized travel
Most travel models in the U.S. do not consider travel by non-motorized modes such as walking
and bicycling. The major outputs of models typically include traffic volume and speeds along
major roads. Many also include ridership on transit systems. Non-motorized travel historically
has played a much smaller role if considered at all. This is due in large part to the difficulty of
obtaining empirical information on the amount of non-motorized travel.
Recently there has been more interest in modeling non-motorized travel. This is particularly
important for this analysis since concentrated infill development is likely to produce more
frequent and shorter trips were walking or bicycling is more feasible. However, among the
three models in this analysis, only Boston's currently considers non-motorized trips.
2.1.3 Trip chaining and tour based modeling
One major criticism of the trip-based model is that many trips are actually part of a larger
journey. Essentially, stops are treated as independent decisions rather than linked to the overall
purpose of the tour. Additionally, while non-home based trips are generated by four step
models, their frequency is based on household characteristics. Overall these structural features
make it difficult to examine how more accessible land use patterns might reduce vehicle travel.
Such changes may lead to more efficient trip-chaining by car or greater use of transit because of
the ability to accomplish tasks on the way to or from the station.
14
-------
In recent years a few urban areas have developed tour based models, which explicitly model the
formation of tours and can be responsive to the effects of land use patterns on trip chaining.
However, none of three partner communities has a tour based model.
2.2 Emissions Analysis Issues
Two issues are particularly important to appropriately estimating emission impacts. As
discussed above, the degree to which travel demand analysis is sensitive to the characteristics
that define good infill development is critical for emissions analysis. Additionally, how the
emissions analysis takes advantage of the additional capabilities incorporated into MOBILE6
model is important. While MOBILE65 was used in all three of the analyses, there were
differences in its application. The details are described under each case in the technical
appendix. A section specifically addressing the benefits of using MOBILE6 rather than
MOBILES is also included. Staff from Cambridge Systematics and Industrial Economics
worked with the participating organizations to describe potential approaches. In particular,
they discussed how current standard procedures use in the region could be enhanced. The
approaches described in the following subsections reflect the results of this technical assistance
and coordination.
2.3 Evaluation Measures
The scenarios analyzed and the analytical procedures used by the three partner communities all
involve the use of regional travel demand models and EPA's MOBILE6 emissions estimation
software. A variety of measures can be obtained from these procedures that are relevant to the
estimation of the effects of brownfield development. These include measures of the amount of
travel, the quality of transportation level of service, and the environmental impacts related to
emissions.
This section describes the evaluation measures used in each partner community. The general
measures are the same for all three communities, but in some cases the agency was unable to
provide some details. In all three cases the measures were computed for the whole region.
Where the scenarios target specific subregions (in Boston and Charlotte), the measures are also
presented for these subareas.
2.3.1 Measures of the Amount of Travel
Person trips - All of the scenarios examined in the three partner communities preserved the
amount of development regionally; any growth moved to a specific area of concentration was
"moved" or reallocated from somewhere else in the future trend scenario. However, under
different development patterns, the total number of person trips could change. In addition, the
amount of development and, therefore, the number of person trips in sub-areas may change.
This occurred in Scenarios 3 and 4 in Charlotte, where some development was assumed to occur
in station areas rather than outlying areas.
5 The initially released version of MOBILE6, now referred to as MOBILE6.0, incorporated
capabilities only for CO, VOC, and NOx emissions. EPA subsequently released draft MOBILE
version 6.2 that incorporates particulate matter and air toxic emissions, and also a draft approach
for estimating carbon dioxide, COa, emissions. The original MOBILE6.0 CO, VOC, and NOX
capabilities remain unchanged in draft MOBILE6.2, and so these results are identical with both
versions.
15
-------
Average trip lengths - By definition, brownfields are located in areas that are already
developed, and therefore developments in such areas will more likely be closer to potential
travelers. Such locations include residences of employees, customers, and visitors to the
developments. Comparing the average person trip lengths among alternative scenarios
provides a means of quantifying the reduction in travel distance associated with developments
that are located in such areas.
Vehicle-miles traveled - The total number of vehicle-miles traveled (VMT) is a standard
measure of the amount of automobile and commercial vehicle travel consumed. Previous work
has shown that infill redevelopment could result in lower VMT on aggregate and per capita
than for a similar amount of development located away from current populations. This is due
not only to the shorter average trip lengths associated with such scenarios, but also to the fact
that brownfield sites are also more likely to be located near public transportation, which means
that travelers to the development are more likely to have choices for travel. It should be noted
that even with regionally lower trip lengths and automobile mode shares, there still may be
increases in VMT in subareas where there is a greater amount of development under
brownfields development scenarios.
Vehicle-hours traveled - The total number of vehicle-hours traveled (VHT) is another standard
measure of the amount of vehicle travel. Brownfield/ infill redevelopment could result in
lower VHT than for a similar amount of greenfields development. This is due not only to the
lower VMT associated with such scenarios, as described above, but also potentially to the
higher average speeds, which are discussed below. It should be noted that even if VMT
declines and speeds increase in the region, there still may be increases in VHT in subareas
where there is a greater amount of development under brownfields development scenarios.
Transit mode shares - The percentage of trips made by transit may change due to changes in
land use patterns or changes in the transit or highway level of service. Comparing transit mode
shares among alternative scenarios provides a way of measuring how well the scenario
provides additional opportunities for persons to travel by means other than private auto.
Walk mode shares - The percentage of trips made by walking (and bicycling) may change due
to changes in land use patterns or changes in the transit or highway level of service. Comparing
walk mode shares among alternative scenarios provides a way of measuring how well the
scenario provides additional opportunities for persons to travel by means other than private
auto. Travel by walking and bicycling is not considered, and therefore cannot be estimated, by
the travel models in Charlotte and Denver.
2.3.2 Measures of Transportation Level of Service
Average speed - If brownfield/infill development reduces VMT, the subsequent reductions in
congestion could increase in vehicle speeds. The average speeds for each scenario are
compared, where available, by roadway functional classification. It should be noted that in sub-
areas where VMT increases under brownfield/infill development scenarios, there could be
associated decreases in speeds. For this project, average speeds for a region or sub-area were
computed as the VMT divided by the VHT.
Congestion - The amount of congestion is measured as the difference in the vehicle-hours
traveled (VHT) under free-flow conditions and the VHT under congested conditions. One of
the benefits of decreased VMT could be a reduction in levels of congestion.
16
-------
2.3.3 Measures of Environmental Impacts
Emissions - The pollutants evaluated in each study are the three major categories computed by
EPA's MOBILE6 model. Travel speeds and the vehicle fleet in a particular region both impact
the emissions per mile of travel. MOBILE6 takes inputs from travel demand models and
produces regional emissions. Outputs are produced for three pollutants:
• Volatile organic compounds (VOC);
• Carbon monoxide (CO); and
Nitrogen oxides (NOx).
3.0 Comparison of Results
This section compares the results from the three case studies in terms of changes in the amount
of travel, congestion and speeds, and emissions. The localized effects of additional
development in specific areas and the potential for mitigation are also examined.
3.1 Land Use and Vehicle Miles Traveled
Since all three case studies dealt with the effects of concentrating growth in specific areas, the
effects of the land use changes on the amount of travel in the three studies can be compared.
Table 3.1 compares the amount of land use change in each case study. The Boston and
Charlotte case studies each have one alternative pattern, while the Denver case study has two.
The overall amount of development relocated in each Denver scenario is approximately the
same, however, the degree of concentration differs (31 versus 10 regional centers) and alters the
travel demand outcomes of the two alternate scenarios.
Table 3.1 - Magnitude of Land Use Change Under Each Scenario
Boston Charlotte Denver
13 Entire
Towns Region
Percentage of Development Moved in Alternative Scenarios
Households
Employment
Average change
14%
22%
17%
1%
4%
2%
4%
1%
3%
7%
9%
10%
The land use changes in Table 3.1 are expressed as the percentages of total households and
employment that are reallocated from the trend scenario under each land use alternative.
Because all of the land use changes in the Boston case study occurred within the 13 towns, it
makes sense to present the Boston results in the context of the "13 Towns" area only. In
Charlotte, development under the South Corridor alternative was shifted from areas growing
under the trend scenario, and it makes sense to present the results as percentages of the entire
region.
17
-------
Table 3.1 summarizes the percentage of each region's jobs and homes relocated in the land use
scenarios. The row labeled "average change" represents a weighted average of the growth rates
for households and employment. The weights represent the relative contributions of
households and employment centers to trip making in the trip generation models.
3.2 Congestion and Speed
Table 3.2 shows the estimated changes in the levels of congestion and average speeds for the
three case studies. The increases in average speeds are modest for all of the case studies. This is
dependent on both the differences among the regions transportation systems and the specific
land use alternatives evaluated. The Denver results indicate that, despite large VMT reductions
in the 10 regional center scenario, congestion is higher and speeds are a bit lower. Although the
31 center scenario does less to reduce overall travel, spreading development out across more
infill locations reduces congestion at the sites. In Charlotte, congestion at the regional level is
significantly reduced by concentrating more development around the rail stations. However, in
the corridor itself, congestion levels do not change much as the increased share of trips made by
transit is offset by a greater share of the region's trips being shifted to the area. In the Boston
analysis, overall travel delays are reduced substantially in the 13 towns, but increase somewhat
at a regional level.
3.3 Emissions
Table 3.2 shows the average emissions rates from the model runs. It is important to recognize
that per mile rate will vary among the communities due to differences in average speeds,
climatic conditions, vehicle fleet mixes and other factors that affect emissions. The rates in
Table 3.3 represent the average rates per mile for the baseline condition; the rates for the other
scenarios are similar to those for the baseline scenario.
Table 3.2 also shows that the emissions rates vary in several ways. The VOC rate in Boston is
about 20 percent lower than in Charlotte and Denver, while Boston's NOx rate is about one
third lower than those in Charlotte and Denver. Charlotte's rate for CO emissions is about half
of the rate in Denver and two thirds of the rate in Boston.
Table 3.3 presents the estimated changes in emissions for the land use scenarios for the three
case studies.
18
-------
Table 3.2 Average Emissions Rates for Each Community (Base Case)
VMT (000' s)
Emissions (kg)
VOC
CO
NOx
Boston*
3,313
783
30,615
661
Charlotte
23,078
6,849
136,263
6,614
Denver
92,308
28,200
1,018,200
28,200
Average Emissions Rate (g/mile)
VOC
CO
NOx
0.236
9.241
0.200
0.297
5.904
0.287
0.370
10.922
0.322
* p.m. peak period and "13 Towns" area only
Table 3.3 Estimated Changes in Emissions for Each Community
Percentage of Development Moved
Change in VOC emissions
Change in CO emissions
Change in NOx emissions
Boston*
17%
-5.5%
-4.8%
-8.1%
Charlotte
3%
-1.4%
-1.2%
-1.1%
Denver
31 Centers
10%
-3.2%
-2.8%
-2.7%
Denver
10 Centers
10%
-4.0%
-3.5%
-3.6%
* "13 Towns" area only
3.4 Effects of Increased Development in Specific Areas
All three of the case studies examined alternative land use patterns where growth would be
concentrated in specific areas. These changes were compared to a baseline consistent with most
recent trends. In effect, this implies that some parts of each region will have more development
in the alternative land use scenarios than in the baseline. The result would be more trips to and
from these areas, and possibly more congestion. The extent to which the increased
development - and the associated trips - lead to more vehicle travel, congestion and emissions,
will depend on the length of car trips and use of alternative modes. It is worthwhile to examine
the types of localized impacts that might be introduced by the alternative land use patterns.
19
-------
Although this type of analysis could be done using the results from any of the three partner
communities, Charlotte is the best suited to illustrating the key issues:
• All of areas with more development in the alternative scenarios are located within a
single corridor.
• As part of their analysis, Charlotte summarized the results for the corridor and
station areas as well as the region.
• One of the scenarios tested included street network improvements designed to
mitigate these localized effects. This was not done in the other two case studies.
All of the redirected development was focused on infill sites close to three stations in the South
Corridor. (More detailed presentation of the analyses are presented in technical appendix C.)
The 16,500 households reallocated in the alternative land use scenario nearly triples the number
of people living near the station areas. However, the 10,500 jobs relocated near the rail stations
increases total employment in the corridor more modestly, by about 50 percent.
Table 3.4 summarizes the model results for both the South Corridor and the Region. The tables
reveal that while travel and emissions do increase in the Corridor, they are substantially less at
a regional level. This result does reveal a tension between the local and regional impacts that
such scenarios can produce. When substantially more development is focused around station
areas there will also be an increase in vehicle travel, congestion and emissions in the local area.
While road improvements
Table 3.5 demonstrates that at least some of the impacts of the increased development in the
Station Areas can be mitigated to a certain extent. While the street network improvements
implemented as part of Scenario 4 would increase VMT in the Station Areas even more than the
increases due to the land use changes (probably by diverting additional through traffic), the
percentage increases in emissions are lower than the percentage increases in VMT. The level of
congestion in the corridor would drop to only six percent higher than the baseline condition,
and average speeds would actually be higher than in the baseline scenario.
20
-------
Table 3.4 Summary of Charlotte Results (Regional Total vs. Station Areas)
Scenario 1 Scenario 2
Scenario 3
Scenario 4
Measure
Baseline
Transit
Development Station Area
Around Road
Transit Improvements
VMT (000' s)
Mecklenberg County
Station Areas
VOC (kg)
Mecklenberg County
Station Areas
CO (kg)
Mecklenberg County
Station Areas
NOx (kg)
Mecklenberg County
Station Areas
Congestion (OOOs of VHT)
Mecklenberg County
Station Areas
Average Speed (mph)
Mecklenberg County
Station Areas
Transit Trips (000 's)
Mecklenberg County
Station Areas
Person Trips (000' s)
Mecklenberg County
Station Areas
Transit Share
Mecklenberg County
Station Areas
23,078
665
6,849
242
136,263
4,496
6,614
207
114
3.40
29.7
25.2
139
2.9
4,156
89
3.4%
3.3%
23,044
658
6,837
240
136,069
4,452
6,605
205
113
3.35
29.8
25.2
145
3.9
4,156
89
3.5%
4.4%
22,604
728
6,742
257
134,451
4,762
6,532
217
109
3.93
29.8
25.0
150
10.0
4,156
185
3.6%
5.4%
22,609
760
6,737
261
134,402
4,932
6,529
225
109
3.60
29.9
26.5
150
10.0
4,156
185
3.6%
5.4%
Table 3.5 Summary of Differences between Baseline and Alternative Scenarios
(Mecklenberg County)
Measure
VMT
VOC
CO
NOx
Congestion
Average Speed
Transit Trips
Transit Share
Transit
Only
-0.1%
-0.2%
-0.1%
-0.1%
-0.9%
0.3%
4.3%
2.9%
Transit Plus
Infill
-2.1%
-1.6%
-1.3%
-1.2%
-4.4%
0.3%
7.9%
5.9%
Infill with
Road Improv.
-2.0%
-1.6%
-1.4%
-1.3%
-4.4%
0.7%
7.9%
5.9%
21
-------
3.5 Conclusions and Implications
The results of this project demonstrate that land use patterns involving infill and the
redevelopment of brownfield areas provide regional transportation and environmental benefits,
specifically reductions in vehicular distances traveled and congestion as well as emissions from
vehicles. This section summarizes the conclusions and discusses the implications of these
conclusions in the context of implementation.
3.5.1 Reductions in Vehicle Travel and Emissions
All three case studies indicated that the amount of regional VMT (aggregate and per capita) can
be lowered by concentrating future development in infill areas, closer to trip destinations and
public transit, as compared to conventional suburban development patterns. Although it is not
possible to make general statements about these reductions based on three specific case studies,
it is clear that for these three regions, concentration of future development into specific areas
results in substantial reductions in regional vehicle-miles traveled (VMT), and that greater
amounts of travel reduction occur if more development is concentrated in these areas. In these
case studies the reductions in VMT within the areas within which development was moved
were in the two to five percent range. Importantly, these reductions in VMT result exclusively
from reductions in trip lengths, not from reductions in trip-making. In other words, travelers
are satisfying their mobility goals with fewer motorized miles.
The reductions in VMT lead to several related benefits. Both vehicle emissions and congestion
are reduced. Emissions reductions in these case studies ranged from three to eight percent
regionally (somewhat less in Charlotte where the land use changes were concentrated in a
single corridor of the region). Overall, congestion decreased by five to seven percent in these
case studies.
The Charlotte case study shows that the amount of vehicle distances traveled (and therefore
congestion and emissions) can be decreased further by coordinating the land use concentration
with good transit service. Because all of the development relative to the base scenario was
moved to light rail station areas in the Charlotte case study, additional opportunities to reduce
vehicle travel existed.
It is important to recognize that while the types of policies analyzed in this project show clear
benefits regionally, those benefits are not uniformly distributed across the region. Areas where
development is concentrated can experience increased total (although lower per capita)
vehicular travel, congestion, and/or emissions, presenting a challenge that may require
additional transportation improvements to mitigate these local impacts. Based on the results of
the Charlotte case study, it was possible to at least partially mitigate these increases through
targeted street network improvements in the areas where development would be concentrated.
3.5.2 Use and Limitations of Travel Demand Models
The state-of-the-practice travel demand models used in the three case study analyses
demonstrate that conventional models are useful in analyzing the effects of brownfield and
infill development on vehicle travel and emissions. This is particularly true when the models
are used to examine alternative land use scenarios; for the most part, transportation planning
22
-------
studies in urban areas assume only a single fixed land use scenario. However, there are several
limitations with conventional modeling. These include the following:
• The necessity of using aggregate analysis zones, which do not allow the analysis of
different land use patterns within a zone, and limit the ability to analyze intrazonal
trips.
• The lack of specific variables to examine the effects of higher density mixed use
developments.
• The lack of ability to analyze non-motorized travel (except in the Boston analysis),
which would likely increase under scenarios with more concentrated land use
patterns.
• The lack of ability to model trip chaining, which might increase under scenarios with
more concentrated land use patterns.
The effect of these limitations, though, can be minimized through the implementation of one or
more enhancements, as demonstrated in each of the three pilot communities. These include the
use of smaller zone sizes, consideration of non-motorized travel, use of intrazonal VMT
adjustment factors, and calculation of mixed land use indices. In addition, the use of a trip-
based emissions estimation procedure, rather than one based simply on VMT changes, captures
emission impacts associated with the changes in trip length that occur with brownfield and
infill development strategies.
While even enhanced conventional state-of-the-practice model systems may not be inherently as
good as newer and more specialized alternatives, they nonetheless represent the capabilities
that currently are used by nearly all MPOs and state DOTs. Consequently, they represent a
good presently available approach for evaluating the impacts of a mix of land use and
transportation strategies. Based on the experience of working with these three pilot
communities, the conclusion is that application of these methodologies sufficiently captures the
impacts of regional brownfield and infill policies to permit the results to be incorporated into
transportation and air quality decision-making.
3.5.3 Relative Contribution of Land Use Strategies
Compared to other air pollution control and transportation strategies, land use policies appear
to offer a way to reduce emissions. For example, in Boston the estimated 239,000 mile reduction
in vehicle travel associated with the concentrated development scenario is greater than the
reductions estimated for most of the transit projects outlined in the Regional Transportation
Plan.6 While this represents less than one percent of the mobile source VOC, NOx, and CO
emissions budgets for the entire region, it reflects only land use changes in 13 of the 101 cities
and towns in the region.
5 Regional Transportation Plan 2004-2005 of the Boston Region MPO, Prepared by the Central
Transportation Planning Staff for the Boston Metropolitan Planning Organization, September 11,
2003.
23
-------
3.5.4 Future Research Needs
As discussed throughout this report, the currently available models make several
simplifications about changes in travel behavior. Additional research is needed to expand the
application of four-step travel models to evaluate urban infill development. Specifically,
enhancements are needed to improve methodologies for estimating non-motorized trips and
accounting for changes in the number and average trip length as a function of changes in the
mix of development and density. Other areas of future research that would be useful in
validating the results of these case studies include trip generation rate changes, changes in trip
balancing procedures, and trip table factoring. Furthermore, development of a mixed-use index
based on empirical research would also be valuable.
24
-------
Sources Cited
Boston Metropolitan Planning Organization, 2007, The Recommended Transportation Plan, Ch
13 in Journey to 2030, Central Transportation Planning Staff.
http://www.bostonmpo.org/bostonmpo/resources/plan/2030plan-13-15.pdf
Ewing, R. and R. Cervero. 2001. Travel and the Built Environment: A Synthesis. Transportation
Research Record 1780, 87-114.
Federal Highway Administration, 2004, Emissions Benefits of Land Use Planning Strategies
FHWA TOPR 29, prepared by The Louis Berger Group, Inc..
http://www.fhwa. dot, gov/environment/conformity/benefits/index. htm
Federal Highway Administration, 2006, Comparative Cost Effectiveness of Potential CMAQ
Projects, Appendix 4 in Interim Program Guidance: The Congestion Mitigation and Air Quality
(CMAQ) Improvement Program.
http://www.fhwa. dot. gov/environment/cmaqpgs/06guide. htm# Appendix4
Transit Cooperative Research Program Land Use and Site Design: Traveler Response to
Transportation System Changes, Report 95 Chapter 15, Transportation Research Board,
Washington DC, 2003.
Transportation Research Board, 2005, Does the Build Environment Influence Physical
Activity? Examining the Evidence, Committee on Physical Activity, Health,
Transportation and Land Use. TRB Special Report 282, Washington, D.C.
U.S. Department of Transportation, Travel Model Improvement Program (TMIP),
http ://tmip. fhwa. dot, gov/
U.S. EPA, Comparing Methodologies to Assess Transportation and Air Quality Impacts of
Brownfield and Infill Development, EP A-231 -R-01 -001.
http://www.epa.gov/smartgrowth/pdf/comparing_methodologies.pdf
U.S. EPA, Granting Air Quality Credit for Land Use Measures: Policy Options, EPA SR99-09-
01. http://www.epa.gov/otaq/stateresources/policy/transp/landuse/lupol.pdf
U.S. EPA, The Transportation and Environmental Impacts of Infill Versus Greenfield
Development: A Comparative Case Study Analysis, EPA publication number 231-R-99-005.
25
-------
Detailed Technical Appendices
Appendix A - Boston
Background
The Metropolitan Area Planning Council (MAPC) and the staff of the Boston MPO tested the
impact of different development patterns in 13 communities in an arc along the 1-495 corridor.
Some of the questions the community wanted to answer include the following:
• Should development be in community centers or along 1-495?
• What is the impact of mixed used development on future transportation
infrastructure?
The project ran two scenarios - one based on current development patterns, the other based on
concentrating development in town centers or 1-495 interchanges. Both scenarios projected the
same amount of employment and household growth, with the concentrated growth scenario
moving about 14 percent of households and 22 percent of employees into more compact
developments.
The traditional four-step model was modified by disaggregating the existing 54 zones into 137
zones, and also used the MOBILE6 model to calculate emissions estimates. It focused on peak
travel and assumed the same transportation and highway improvements for each scenario.
The transportation and air quality impacts of current development patterns were compared to
those that could result from an alternative land use scenario. The analysis quantified
congestion impacts, vehicle miles traveled (VMT), vehicle delays, and commuter rail use.
The 13 communities in the study area were chosen according to three criteria:
• They are in the 1-495 corridor, which has the highest growth rate and largest amount
of developable land in the region.
• They represent the three subregions in the corridor.
• They include the area considering formation of a new transit authority.
The 13 towns are projected to grow to a total population of 334,681 over the next 25 years. This
represents a projected 127,174 households. The forecasted employment for 2025 is 465,792.
There was no attempt by MAPC and CTPS to balance the growth in households and
employment. It turns out that the trend scenario assumed a higher growth rate for employment
(about 35 percent) than for households (about 15 percent). This resulted in the population
estimate being close to the 2025 MAPC forecast, but the employment being higher than the
forecast. The imbalance was not reconciled, and the transportation model had to account for
the imbalance by assuming a higher rate of trips produced outside the MAPC region attracted
to communities within the region, including the 13 communities in this study.
The concept of trend was introduced by MAPC to determine the effects of the maximum
amount of development allowable under current regulations. While trend is not necessarily
associated with any particular forecast year, the analyses performed for this project was
generally assumed the towns would develop available parcels by 2025. Buildout analyses have
A-l
-------
been done for most of the MAPC communities. This pilot study was intended to determine the
effects of alternative development patterns on transportation. Although an additional scenario
using projections based on historical development patterns would have been useful, resource
constraints did not allow for this analysis.
Two scenarios were analyzed. Thee trend scenario, representing the maximum allowable
development under current zoning. Scenario 2 concentrated development in existing town
centers or near 1-495 interchanges. MAPC staff developed this scenario manually by selecting
available infill sites that could accommodate the redirected growth. Examples of the alternative
sites in scenario 2 included the Ashland Center brownfield site, infill development in
Hopkinton Center, and additional development in the western section of Hudson. Scenario 2
also included some redevelopment of sites with a current active land use. The amount of
employment and households by type was held constant between the two scenarios for each
community. Under Scenario 2 about 18,000 (14 percent) of the households and about 102,000
(22 percent) employees in the 13 towns were moved relative to Scenario 1.
Detailed Results
Tables A.I and A.2 present the results for two geographic areas. The first area is the entire
metropolitan Boston region as modeled by the Metropolitan Area Planning Council (MAPC)
and the Central Transportation Planning Staff (CTPS). The second area is the "13 Towns" area,
which consists of those towns where development was redistributed under Scenario 2. When
reviewing these results, it should be noted that while growth is redistributed, the total amount
of development within the 13 Towns area is held constant for both scenarios. The analysis was
performed only for the p.m. peak period. Information on person, transit, and walk trips and on
average trip lengths was provided only for the 13 Towns area.
According to the results shown in Table A.2 for the 13 towns, the land use pattern under
Scenario 2 would result in significant reductions in VMT, emissions, and congestion, on the
order of five to 10 percent, while speeds would increase. Regionally, the reductions in VMT,
emissions, and congestion would be on the order of 0.5 to 1.6 percent. This is quite large, given
that the 13 Towns area accounts for only 10 percent of regional travel.
Within the 13 towns, 14 percent of the households and 25 percent of the employment were
relocated under Scenario 2. This represented about half of the growth in households and about
15 percent of the growth in employment from the base year to 2025.
A-2
-------
Table A.l. Analysis Results for Boston (p.m. peak period)
Measure
VMT (000' s)
VHT (000' s)
VOC (kg)
CO(kg)
NOx (kg)
Congestion (OOOs of
VHT)
Average Speed (mph)
Person Trips (000' s)
Transit Share
Walk Share
Avg. Trip Length (miles)
Area
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Scenario 1
Trend
30,577
3,313
1,315
154
6,329
783
294,836
30,615
5,871
661
565.41
77.93
23.20
21.50
n/a
654
n/a
0.22%
n/a
6.21%
n/a
14.95
Scenario 2
Focused Development
30,338
3,159
1,300
145
6,295
740
293,089
29,149
5,832
608
556.59
72.29
23.34
21.76
n/a
654
n/a
0.23%
n/a
7.37%
n/a
14.30
Scenario Definitions:
1. Base - trend for all towns
2. Concentrated redevelopment in 13 towns
It can be concluded from the Boston analysis that the redistribution of development in
concentrated areas has significant benefits in terms of reduced vehicular travel, decreases in
emissions, and improved congestion levels. The reduced vehicular travel is a result of shorter
trip lengths and mode shifts from auto to transit and walking. The benefits of having fewer and
shorter trips are noticeable not only in the areas where the development is redistributed, but
also regionally. Within the 13 Towns area the reductions in VMT, congestion, and emissions
and the increase in transit share are all on the order of five to eight percent. Regionally, these
effects are on the order of one to two percent although it should be noted that in most cases the
benefits extend beyond the boundaries of the 13 towns.
A-3
-------
Table A.2. Difference between Boston Scenarios
Measure
Scenario 2
Area Change Due to
Value
Minus Scenario 1
Focused Development
Percent
VMT (000' s)
VHT (000' s)
VOC (kg)
CO (kg)
NOx (kg)
Congestion (OOOs of VHT)
Average Speed (mph)
Person Trips
Transit Share
Walk Share
Avg. Trip Length (miles)
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
Entire Model
13 Towns
-239
-154
-15
-9
-34
-43
-1,747
-1,466
-39
-53
-8.82
-5.65
0.14
0.26
n/a
-475
n/a
0.01%
n/a
1.16%
n/a
-0.65
-0.8%
-4.7%
-1.2%
-6.2%
-0.5%
-5.5%
-0.6%
-4.8%
-0.7%
-8.1%
-1.6%
-7.2%
0.6%
1.2%
n/a
-0.1%
n/a
4.2%
n/a
18.8%
n/a
-4.3%
Scenario Definitions:
1. Base - trend for all towns
2. Concentrated redevelopment in 13 towns
Transportation Analysis Methods
CTPS performed the transportation and emissions modeling for the two scenarios described
below. The transportation modeling was performed using the regional travel model
maintained by CTPS. This is a conventional four-step travel model based on person trips —
including both motorized and non-motorized trips. Trip generation, trip distribution, mode
choice, and highway and transit assignment were performed using the EMME/2 modeling
software. Emissions modeling was performed using the most recent version of EPA's MOBILE
program.
To explain the differences between the two land use scenarios, a greater level of detail was
needed within the 13 communities. The zone structure within these communities was
disaggregated from 54 zones to 137 zones. The highway and transit networks were assumed to
be the same for both scenarios, and the 2025 regional transportation plan networks were used.
No new transit or highway improvements were assumed in the corridor under either scenario.
A-4
-------
The model runs were performed for the two peak periods (a.m. and p.m.) only. Neither
transportation nor emissions modeling for the entire day was performed. The time of day
modeling was performed by factoring the daily trips after trip generation into peak and off-
peak trips, and performing trip distribution separately for these two time periods, factoring the
peak trips into a.m. and p.m. and the off-peak trips into mid-day and night, and then
performing mode choice, and assignment separately for each of the four time periods. The
results were prepared only for the p.m. peak period
Intrazonal VMT Adjustment
The mode choice model embedded in the CTPS regional travel model splits the intrazonal trips
that are output from the trip distribution model into auto and transit person trips and non-
motorized (mostly walk) trips.
One of the inputs required by the model to split the intrazonal trips is the intrazonal
impedances by the walk and auto modes. The intrazonal impedance for either the auto or walk
mode for any zone is estimated as 50% of the impedance by that mode to the nearest zone.
These impedances are usually held constant among different scenarios. Therefore, any
reduction in intrazonal impedance resulting from a "Smart Growth" strategy (more mixed use)
is not reflected in the CTPS model.
It has been hypothesized that the CTPS model may be underestimating the number of
intrazonal trips in Scenario 2. It would make sense that the more "mixed-use" a development
is, the higher the percentage of intrazonal trips.1 One way to measure the mixed-use nature of a
zone is to combine the different employment types plus housing. To address the potential
under representation of intrazonal trips, CTPS developed a procedure that uses the concept of
"mixed use index"2 and corrects for the presumed underestimation of intrazonal trips. In this
procedure, when more intrazonal trips are added to a given zone, an equivalent number of
interzonal trips to and from that zone are removed.
The mixed use index for each zone is defined as follows:
BE E x BH H
MUI=
BE E + BH H
where:
MUI = mixed use index for the zone
E = employment in the zone
BE = weight for employment
H = number of households in the zone
BH= weight for households
1 For example, in Seattle, neighborhoods with mixed use development are almost four times as
likely to be able to meet trip needs within a mile of the home, compared to surrounding areas. For
more information, see http://onlinepubs.trb.org/onlinepubs/tcrp/tcrp rpt 95cl5.pdf.
2 Rossi, Thomas. "Potential Model Enhancements for EPA Project," Memorandum to Erik Sabina,
DRCOG, November 15, 2002.
A-5
-------
This formula will produce higher values when either the amount of development increases or
its mix of residential and commercial uses becomes more even. The weights are used to
represent the relative importance of employment or households in trip making.
The calculations embedded in the procedure are illustrated in the following steps. All
calculations were performed for each zone in the study area. The procedure was implemented
in an Excel spreadsheet.
1. The weight for employment for each zone, BE, was determined by dividing the total
employment into basic, retail and service types, applying different trip generation
rates for each employment type (based on the CTPS trip attraction model), and
estimating a weighted average trip generation rate.
2. The weight for households for each zone, BH, was assumed to be 10, which is the
typical trip production rate per household.
3. Using the employment and household weight factors and the land use/
demographic data, the value of MUI for each zone was calculated for both
scenarios.
4. Next, for each zone, the Intrazonal Trip Adjustment Factor (ITAF) was computed
as the ratio of the MUI for the Scenario 1 to the MUI for the Buildout Scenario.
5. For each zone, the number of intrazonal trips obtained from the trip distribution
model for Scenario 1 was multiplied by the ITAF for the zone. The resulting
number is the adjusted intrazonal trips for Scenario 1.
6. The difference between the model-estimated and adjusted intrazonal trips for each
zone represents the change in intrazonal trips resulting from a better mixed-use
type of development in the Smart Growth scenario. Since the total number of trips
in the study area is largely unaltered (population and employment do not change),
the increase in intrazonal trips should be accompanied by an equal decrease in
interzonal trips.
7. It was assumed that 20 percent of the new intrazonal trips would divert to the walk
mode. The remaining 80 percent would be auto trips. CTPS based this assumption
on experience in other parts of the region.3
8. The VMT resulting from the additional intrazonal auto trips was estimated by
multiplying the number of new intrazonal auto trips by the average intrazonal trips
length for the zone.
9. The average trip length for all interzonal trips originating or destined to each zone
was estimated from the trip tables.
3 Although it is difficult to isolate the effect of mixed use development from density and other
confounding factors, several studies have been conducted that support the substitution of walking
for motorized transit as mixed uses increase. For example, traveler response studies in Houston
and Seattle were within this range. See
http://onlinepubs.trb.org/onlinepubs/tcrp/tcrp rpt 95cl5.pdf for more information.
A-6
-------
10. For each zone, the average interzonal trip length estimated in Step 9 was multiplied
by the decrease in interzonal trips to estimate the reduction in VMT from these
trips.
11. The net VMT reduction was calculated by adding the VMT increase from Step 8
and VMT decrease from Step 10.
Emissions Analysis Methods
Subsequent to completion of the initial MOBILES-based analysis for the 1-495 corridor, the
Massachusetts DEP completed development of 2007 MOBILE6 files for the SIP development
and also developed generic files for use by others in conducting build/no-build and other
similar emission analyses. The files are set up to produce emission factors by 1.0 mph
increments for freeways and arterials, which then can be applied in a spreadsheet to link-level
VMT by speed, similar to the MOBILES approach.
Based on the availability of these new MOBILE6 results and the successful experience of the
Massachusetts Highway Department in using these data, CTPS applied MOBILE6 for purposes
of the 1-495 corridor EPA brownfield and infill development analysis.
A-7
-------
Appendix B - Charlotte
Background
The City of Charlotte modeled transportation impacts from alternative land use scenarios in a
corridor proposed to be served by a new light rail line. VMT and emissions estimates for the
alternative scenarios were compared to those for the 2025 Regional Transportation Plan, which
assumes continuing trends in land use. The scenarios analyzed include land use shifts to
proposed transit (including light rail) station locations. The regional transportation model
maintained by the City was used to estimate the effects of these land use scenarios. The project
also examined alternative transit and network capacity improvements.
Charlotte's analyses focus on the south corridor, where a light rail line has been proposed. A V2
cent sales tax has been passed to fund the development of light rail. The build and baseline
scenario for this light rail line have already been defined in support of the city's application to
the Federal Transit Administration (FTA) for New Starts funding. To conform to FTA
requirements, the land use for both scenarios was assumed to be the same. This land use
scenario was based on existing trends. Figure 2.2 shows the locations of the transit corridors in
Charlotte's plan, and Figure 2.3 shows the south corridor in greater detail.
The City was interested in examining the effects of different land use patterns associated with
transit stations. Station area plans were developed for seven stations. These plans included
both a maximum and a minimum estimate of potential development. The latter was assumed
for the scenarios due to the fact that it was felt to be more reasonable and was already the basis
of previous transportation model runs.
The following alternative scenarios were modeled by the City of Charlotte:
• Scenario 1 - Transit No Build (as defined for FTA New Starts analysis) - with trend-
based land use assumptions;
• Scenario 2 - Transit Build Scenario (as defined for FTA New Starts analysis) - with
trend-based land use assumptions; and
• Scenario 3 - Transit Build Scenario with Revised Land Use - with land use
assumptions reflecting development plans at stations; and
• Scenario 4 - Transit Build Scenario with Revised Land Use and Highway
Improvements - with land use assumptions reflecting development plans at stations,
with street network improvements to address issues resulting from the revised land
use.
Scenario 1 is the baseline. Scenario 2 adds the proposed light rail line to the baseline. Therefore,
the differences between Scenarios 1 and 2 can be considered to be due to the introduction of the
new transit service. Scenario 3 introduces the alternative land use pattern, including the station
area plans, to Scenario 2. This land use plan includes the redevelopment of some brownfield
sites. The differences between Scenarios 2 and 3 are due to the land use changes in the presence
of the improved transit. Scenario 4 adds the street network improvements to Scenario 3. The
B-l
-------
differences between Scenarios 3 and 4 can be considered to be due to the highway
improvements in the presence of the improved transit and revised land use.
These scenarios comprise two different transit scenarios (scenarios 1 and 2), two different land
use scenarios (2 and 3), and two different highway network scenarios (3 and 4), in each case
with everything else held constant.
Both land use scenarios assumed the same amount of growth and total development in the
county, however the alternative land use scenarios attract growth that would have occurred in
parts of the county outside the study area under the trend scenario. By the analysis year of
2025, it is projected that the employment in Mecklenburg County, where Charlotte is located,
will reach 764,862. The forecasted 2025 population is 944,649, representing 397,001 households.
Under the alternative land use scenarios (3 and 4), a total of 16,500 households (about 4.2
percent of the county total) and 10,500 employees (about 1.4 percent of the county total) are
located in the south corridor that under the trend scenario would have been located elsewhere
in the county.
Detailed Results
The City of Charlotte has produced a detailed documentation report for the analysis they
performed for Charlotte4. While it is impossible to incorporate all of the details provided in that
report here, the report is referenced where appropriate.
Table B.I presents the analysis results for Charlotte. The emissions analysis was performed in
two ways: 1) VMT-based only, and 2) including the VMT, vehicle trip, and vehicle based
emissions. To be consistent with the analysis performed for the other two partner communities,
the emissions shown B.I are the VMT-based emissions.
The results for Charlotte are presented for three geographic areas. The first area is the whole of
Mecklenburg County. The second area is the "South Corridor," which is wide enough to
include the location of the transit right-of-way and an interstate or freeway also within the
corridor. The third area, referred to as the "Station Areas," is the most focused on the station
areas. Figure B.I shows the South Corridor (orange) and Station Areas (red).
4 City of Charlotte. "Air Quality Benefits of Brownfields Development: Methodology Report and
Summary Results" (Draft). June 21, 2003.
B-2
-------
Figure B.I. Location of Charlotte South Corridor and Station Areas
TAZ regions
region
• CBD
• S:a1ion Area
Q South Com cor
• Other Csrrittcrs
I I Inner wedge (unchanged)
| | Outer wedge (Donor area)
• Ciraide Mecklenburg
B-3
-------
Table B.I. Analysis Results for Charlotte (weekday)
Measure
VMT (000 's)
VHT (000 's)
VOC (kg)
CO (kg)
NOx (kg)
Congestion
(OOOs of VHT)
Average Speed
(mph)
Transit Trips*
(000 's)
Person Trips*
(000 's)
Transit Share*
Avg. Trip
Length (mi)
Area
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
South Corridor
Station Areas
Mecklenburg Co.
Scenario 1
Baseline
23,078
2,911
665
776
113
26
6,849
1,392
242
136,263
27,199
4,496
6,614
1,364
207
114
17
3
29.7
25.7
25.2
139
9
3
4,156
327
89
3.4%
2.8%
3.3%
8.00
Scenario 2
Transit
23,044
2,892
658
774
112
26
6,837
1,384
240
136,069
27,050
4,452
6,605
1,357
205
113
17
3
29.8
25.7
25.2
145
12
4
4,156
327
89
3.5%
3.6%
4.4%
8.00
Scenario 3
Development
Around
Transit
22,604
2,968
728
758
116
29
6,742
1,400
257
134,451
27,348
4,762
6,532
1,369
217
109
17
4
29.8
25.6
25.0
150
18
10
4,156
423
185
3.6%
4.3%
5.4%
7.93
Scenario 4
Station Area
Road
Improvements
22,609
2,979
760
756
114
29
6,737
1,397
261
134,402
27,376
4,932
6,529
1,370
225
109
17
4
29.9
26.0
26.5
150
18
10
4,156
423
185
3.6%
4.3%
5.4%
7.93
* Trips originating in specified area.
Scenario Definitions:
1. Transit no build scenario
2. Transit build scenario with trend-based land use assumptions
3. Transit build scenario with more focused redevelopment around stations
4. Transit build scenario with revised land use assumptions and station area road improvements
B-4
-------
In examining the results, it is important to recognize that while the total amount of
development is held constant on a regional basis, the amount of development in the South
Corridor and Station Areas increases under the revised land use (Scenarios 3 and 4). For
comparison purposes, the number of person trips generated in each scenario is presented as a
measure of the amount of the increased development in these areas under the revised land use
scenario. The number of person trips increases by 29 percent in the South Corridor area and by
108 percent in the Station Areas under the revised land use scenario.
The number of households reallocated in Scenarios 3 and 4 is 16,500, which is about four
percent of the regional total households. This household reallocation nearly triples the number
of households in the Station Areas and increases the number within the larger South Corridor
area by about 60 percent. The total employment reallocated in Scenarios 3 and 4 is 10,500,
which is about 1.4 percent of the regional total employment. This reallocation increases the
total employment in the Station Areas by about 50 percent and increases the number within the
larger South Corridor area by about 10 percent.
Examination of the differences between Scenarios 2 and 1 shows that the introduction of the
light rail line would result in decreases of about one percent in VMT, emissions of VOC, CO,
and NOx, and congestion in the South Corridor and Station Areas. Speeds would increase
slightly while the transit share would increase by approximately 30 percent. Regionally, there
would be small reductions in VMT, emissions, and congestion along with a 4.3 percent increase
in transit share. Examination of the differences between Scenarios 3 and 2 shows that in the
South Corridor, the revised land use would result in increases in VMT (about three percent),
emissions (one percent), and congestion (four percent) and decreases in speed (0.3 percent) that
more than offset the benefits resulting from the introduction of the improved transit. These
changes are due to the significant increase in development in the area. This indicates that
although the new development would be concentrated in the areas of the new light rail stations,
many of the trips to and from these developments would be made by auto.
However, the percentage increases in VMT, emissions, and congestion are much lower than the
increases in the number of person trips. This reflects the fact that the VMT and emissions
associated with trips through the corridor, which are included in the summaries, would not
change significantly since they would not travel to or from the new development. This also
reflects that the new station area development would generate relatively fewer auto trips, as
indicated by the increases in transit shares in the area beyond the increases associated with the
introduction of the transit improvements alone, as well as shorter trips, due to the more
compact nature of the development in the station areas.
Despite the increases in VMT, emissions, and congestion in the South Corridor, it must be noted
that regionally there would be decreases in these measures. Regionally, VMT and emissions
would decrease by nearly two percent while congestion would decline by nearly four percent.
The highway improvements associated with Scenario 4 would increase VMT in the Station
Areas by 4.3 percent. This is presumably due to traffic that might be diverted to the expanded
roadways since regional VMT (and emissions) would be essentially unchanged. However, the
highway improvements would reduce the level of congestion, even in the Station Areas where
the VMT would increase. The level of congestion would improve in the South Corridor area to
about the same level as in the baseline scenario (0.2 percent higher) even with 29 percent more
B-5
-------
development. Regionally, congestion levels would be nearly five percent better than in the
baseline scenario.
It can be concluded from the analysis that the combination of increased transit service and
corresponding supportive changes in land use patterns can result in significant benefits in terms
of reduced vehicular travel, increased transit use, decreased emissions, and improved
congestion levels. These benefits are regional in nature as developments occur in the transit
corridor rather than in areas where people are more likely to make more and longer vehicle
trips. While vehicle travel and emissions are likely to increase in the areas in the transit corridor
where development is concentrated, these effects can be at least partially mitigated through
selective highway improvements.
Transportation Analysis Methods
The regional land use and travel models maintained by the City of Charlotte were used to
perform the analyses. These models are documented by the City in a separate report5.
The City has a land use allocation model. It assumes regional employment and housing totals
are held constant. The model allows specific developments and proposed projects to be added
manually. There is a subregional component to this model. This model was used to produce
the land use scenario for Scenarios 1 and 2 (the same land use assumptions were used for these
two scenarios). The land use scenario used for Scenarios 3 and 4 was based on this land use
scenario and manually revised according to the station area plans.
The regional transportation model is a conventional four-step model based on motorized trips
(auto and transit). Trip generation, trip distribution, mode choice, and highway and transit
assignment were performed using the TransCAD software. There is also an auto ownership
model. The highway assignment is done for four time periods, two peak and two off-peak
periods.
5 "Air Quality Benefits of Brownfields Development - Charlotte, NC: Methodology Report and
Summary Results, June 21, 2003.
B-6
-------
Figure B.2 Charlotte Transit Corridors
2025 Household Density
Map Layers
;"J| Counties
Corridor
Transit Comdors
Streetcar
MDMU
2025 HH / Sq.ML
0 to 499
53D to 999
Other
Highways
— Thoroughfare
Onsr
: i -
Miies
B-7
-------
Figure B.3 Charlotte South Corridor
South LRT
Map Layers
Sode
Counties
Corrida
Transit Corridors
Node Selection Sets
Slaticrs
2025 HH / SqM.
0 to 499
EDO :c 399
'OOD Io1969
2 00 D to 49 69
Dire-
Highways
Freeway
Thoroughfare
Other
6 :
B-8
-------
Emissions Analysis Methods
The Charlotte analysis was performed using MOBILE6 input files developed in January
2002 by the North Carolina Department of Transportation. Output emission factors are for
calendar year 2025. The input files include the following locality-specific inputs for
Mecklenburg County:
• Vehicle age distributions;
• VMT mix (percent of VMT by vehicle type);
• Inspection and maintenance (I/M) program;
• Soak distributions;
• Temperature;
• Fuel RVP; and
• Vehicle speed.
For each land use/transit scenario, 18 MOBILE6 scenarios were run, each with an average speed
for six different functional classes and three time periods. Speed was estimated from travel
model output as VMT divided by vehicle-hours traveled (VHT). The time periods included
a.m. peak and p.m. peak (two hours each) and off-peak (remainder of the 24-hour period). The
six different functional classes, along with their corresponding MOBILE6 facility type(s), are
shown in Table 3.1.
To account for changes in vehicle trips versus average trip lengths, grams/start emission factors
were applied separately to the total number of vehicle trips in the study area under each
scenario. (Grams/start were calculated as daily start emissions divided by the default number
of daily starts assumed in MOBILE6). Grams/mile emission factors, net of start emissions, were
then applied to total VMT under each scenario.
EPA has not published guidance on the proper use of MOBILE6 to model trip-based versus
VMT-based emissions. The methodology is straightforward for CO and NOx, for which
MOBILE6 reports separate "start" and "running" emission factors. For VOC, however, there
are eight different components that must be allocated to trip ends, VMT, or simply the existence
of the vehicle. Discussions with EPA staff suggested that the following allocation approach was
reasonable:
B-9
-------
Table B.I Functional Classes in the Charlotte Emissions Analysis
Functional Class
MOBILE Facility Type
Other Comments
Urban Interstate
Urban Freeway
Urban Principal Arterial
Urban Minor Arterial
Urban Collector
Urban Local Street
Freeway
Ramp (% from assignment)
Freeway
Ramp (3%)
Arterial
Arterial
Local
Local
No Cold Starts
1. Running - VMT
2. Start - Trip-End
3. Hot Soak Loss — Trip-End
4. Diurnal Loss — Vehicle
5. Resting Loss — Vehicle
6. Running Loss - VMT
7. Crankcase Loss - VMT
8. Refueling Loss - VMT
Diurnal and resting emissions depend primarily on whether the vehicle exists, not how
much it is used. Although there is evidence that proximity to transit is associated with
lower automobile ownership rates, this relationship is complex and is not captured in the
model. Some of these eight components could be affected in other ways by the number of
starts and/or VMT per day, but some effects will be negative rather than positive. For
example, reducing VMT per vehicle may slightly increase diurnal emissions because of the
longer time not running and fewer interrupted diurnals.
Because the emissions analysis performed by the other two partner communities included
only the VMT-based emissions from MOBILE6, the emissions results for Charlotte were
reported in two ways: 1) VMT-based only, and 2) including the VMT, vehicle trip, and
vehicle based emissions.
Table 3.2 shows the percentage of emissions allocated to the VMT, vehicle trip, and vehicle
for a Charlotte arterial at an average speed of 30 mph in 2025.
B-10
-------
Table B.2. Sample Allocation of Emissions to VMT, Trip-Ends, and Vehicles
VMT Portion
Trip-End Portion
Vehicle Portion
voc
LDGV
58%
34%
8%
LDGT
55%
37%
7%
CO
LDGV
61%
39%
0%
LDGT
61%
39%
0%
NOx
LDGV
87%
13%
0%
LDGT
82%
17%
0%
Percentages may not add to 100% due to rounding.
B-ll
-------
Appendix C - Denver
Background
The Denver Regional Council of Governments (DRCOG), which both provides regional
planning services and acts as the MPO for the Denver region, is in the process of extending
their Metro Vision 2020 Plan to a new horizon year, 2030. One of the core elements of the
Metro Vision Plan is the establishment of urban centers around the region, as a means of
focusing development reducing the prevalence of dispersed development, and improving
transit accessibility. As part of the evaluation of the effectiveness of such a measure, it is
important to forecast how different scenarios of infill and brownfield development affect
both transportation and air quality. To identify those impacts, DRCOG carried out the
following activities:
• Defined three alternative development scenarios that reflect varying urban densities
and rates of infill, and forecast households and employment by traffic analysis zone.
• With the assistance of the City and County of Denver's Department of Health and
Hospitals, identified brownfield sites (consistent with EPA's brownfield definition).
Highlight and compared the growth predicted in each of the scenarios between 2002
and 2030 for these brownfield sites.
• Used the land development information from the three scenarios as demographic
inputs for the transportation model.
• Used this information in the air quality model to predict emissions for major
pollutants such as carbon monoxide and nitrogen oxides.
The 2030 population forecast for the DRCOG region is about 3.5 million, which represents
about 1.4 million households. The employment forecast is nearly 2 million.
The types of land use scenarios were defined by DRCOG in cooperation with EPA and the
consultant team. The following is DRCOG's summary descriptions of the three scenarios:
Scenario 1 - Limited Brownfield Redevelopment (Baseline): Under this scenario, the
added growth is spread outside the existing developed urban area. Household growth
followed an increased suburbanization pattern, with new employment located in
suburban employment centers. These employment centers tend to locate at major
interchanges of the freeway system in a metropolitan area. The land use pattern can vary
from high density, office towers (as is often seen around airports) to campus-style office
clusters. The residential patterns tend to follow the suburban development patterns, with
large tracts of residential areas supported by neighborhood retail. This scenario was
developed to provide a generalized base case, replicating development patterns of the last
twenty to forty years. The development pattern for this scenario is shown in Figure C.I
Scenario 2 - Multiple Brownfield Centers: This scenario focuses growth into 31 specified
urban density centers. These centers are spread throughout the metropolitan area, but
tend to concentrate in the urban core of the metro area. Both housing and employment
development is focused into these areas, allowing for increased multi-modal usage (transit,
C-l
-------
pedestrian and biking alternatives). Many of the brownfield development projects
propose a mixed-use emphasis, relying both on connections within the development along
with connections to existing community areas. In many cases, the infrastructure is already
in place (water, power, and transportation) allowing for some costs savings for this type of
development (to offset land prices and clean-up costs). This scenario provides for many,
smaller scale redevelopment projects, which may be a means to overcome costly
infrastructure expenses (matching closer to existing scale) and traditional concerns. The
development pattern for this scenario is shown in Figure C.2
Scenario 3 - Concentrated Brownfield Centers: The third scenario focuses development
into 10 specified redevelopment centers. The scale of the brownfield development is much
higher, in part to represent the scale that may be associated with higher land and clean-up
costs. While the associated development has much higher densities, the projects may
serve as catalyst for further redevelopment in the surrounding community. In the Denver
region, examples of this type of project include the Central Platte Valley, Gates Rubber,
Stapleton, and Fitzsimons sites. These developments have varied shares of residential and
employment components, and still rely on existing infrastructure. However in every case,
the development will likely cause some additional infrastructure improvement to
adequately serve the scale of development proposed. The development pattern for this
scenario is shown in Figure C.3.
The amount of development reallocated into centers was similar under both Scenarios 2
and 3, measured relative to Scenario 1. Under each scenario, about 100,000 households
(about seven percent of the total) and about 180,000 to 190,000 employees (about 14
percent of the total) shifted locations.
Table C.I
Urban Density
Forecast of
Employment
Forecast of
Households
Infrastructure
Requirements
Scenario 1:
Limited Brownfield
Variable
Suburban employment
centers focused at
major freeway
interchanges
Suburban residential
areas near employment
centers
Requires new
infrastructure
Scenario 2:
Multiple Centers
Medium
Focused on 3 1
areas in urban core
Focused on 3 1
areas in urban core
Generally
preexisting
Scenario 3:
Concentrated Centers
High
Focused in 10
redevelopment centers
Focused in 10
redevelopment centers
Preexisting, but
improvements likely
needed
C-2
-------
EPA Brownfield Study
Forecasted Employment
Suburbanization Scenario
Less Than 500
501 -1,500
1,501-3.500
3,501 - 5,000
Over 5,000
Figure C.I Denver Baseline Development Scenario
C-3
-------
EPA Brownfield Study
Forecasted Employment
Multi-Center Scenario
Less Than 500
501 -1,500
1,501 -3,500
3,501 -5,000
Over 5,000
18
20 P*k
21
23 Shelley Lake
33 stapletwi Retteveicipmgru (3)
24 '"• I i-^". i:-
36 Tech Center Ar&pahoe
26 Tof
37 Thornton P»kway
29 Thamton
39 Westminster
30 Westminster Promenade
31 WesTm mstar Ptara
6 BfOGrtifiald Town Canter
TCBD
8 Chflrry Creek
9 Eftglewood Town Center
Figure C.2 Denver 31 Regional Center Development Scenario
C-4
-------
EPA Brownfield Study
Forecasted Employment
10 Center Scenario
Less Than 500
501 -1,500
1,501 -3,500
3,501 -5,000
Over 5,000
Aurora City Centre
Boulder
Broom field
Englewood Town Center
Gates Redevelopment
8 Lakewood
9 Longman!
10 Stapleton Redevelopment
Figure C.3 Denver 10 Regional Center Development Scenario
Detailed Results
Three types of land use scenarios were defined by DRCOG in cooperation with EPA and the
consultant team. Scenario 1 represents baseline conditions, where household growth followed
an increased suburbanization pattern and employment growth was focused in suburban
employment centers. Scenario 2 focuses growth into 31 specified urban density centers spread
throughout the metropolitan area. Scenario 3 focuses development into 10 specified
redevelopment sites. About seven percent of all households and 14 percent of all employment
were relocated in both the second and third scenarios, compared to the baseline scenario.
Tables C.2 and C.3 present the analysis results for Denver. Because the alternative land use
patterns were defined for the entire Denver region, it did not make sense to define subareas for
presentation of the results. They are therefore presented only for the entire region. The results
reflect the trip generation adjustment to the DRCOG travel model described below.
According to the model, concentration of the growth in the 31 centers would result in about a
three percent reduction in regional travel, as measured by VMT. Speeds would, on average,
increase by 0.7 percent, and congestion would decrease by six percent. Emissions of VOC, CO,
and NOx would decline by about three percent.
C-5
-------
The results when growth is concentrated around 10 centers are similar to those for the more
dispersed 31 center scenario. There would be larger decreases in VMT and emissions, but a
smaller decline in congestion and a smaller increase in speed. As with Boston, it can be
concluded from the Denver analysis that the redistribution of development in concentrated
areas has significant benefits in terms of reduced vehicular travel, decreases in emissions, and
improved congestion levels.
Table C.2 Analysis Results for Denver (weekday)
Measure
VMT (000' s)
VHT(OOO's)
VOC (kg)
CO (kg)
NOx (kg)
Congestion (OOOs of VHT)
Average Speed (mph)
Transit Share
Avg. Trip Length
Person Trips (000' s)
Baseline
Scenario
92,308
34,500
1,018,200
30,000
667
31.85
2.4%
6.84
12,695
31 Centers
89,637
33,400
989,900
29,200
625
32.07
2.6%
6.81
12,316
10 Centers
88,966
33,100
982,100
28,900
628
31.98
2.7%
6.78
12,203
Table C.3 Differences between Denver Scenarios
Measure
Change with 31 Centers
Value Percent
Change with 10 Centers
Value Percent
VMT (000' s)
VHT (000' s)
VOC (kg)
CO (kg)
NOx (kg)
Congestion (OOOs of
VHT)
Average Speed (mph)
Transit Share
Avg. Trip Length
-2,761
-1,100
-28,300
-800
-42.27
0.22
0.2%
-0.03
-2.9%
-3.2%
-2.8%
-2.7%
-6.3%
0.7%
9.0%
-0.4%
-3,342
-1,400
-36,100
-1,100
-39.34
0.13
0.3%
-0.06
-3.6%
-4.0%
-3.5%
-3.6%
-5.9%
0.4%
11.5%
-0.8%
Transportation Analysis Methods
The regional transportation model maintained by DRCOG is a conventional four-step model.
Trip generation, trip distribution, mode choice, and highway and transit assignment were
performed using the MinUTP software. The model does have time of day analysis but does not
C-6
-------
include non-motorized travel. The model's parameters are currently being updated, but this
analysis does not reflect the updates.
The current model is not sensitive to detailed microscale-level land use assumptions, mainly
due to its large zone size relative to the scale of typical transit oriented developments. Area
type is an important input into several model components, including (motorized) trip
generation and estimation of roadway capacity. There are currently five area types in the
model: CBD, fringe, urban, suburban, and rural. The same highway and transit networks were
used for all three scenarios.
Intrazonal Trip Changes
Mixed land use development could generate a higher percentage of intrazonal trips than typical
land use patterns. Closer proximity between households and employment sites would facilitate
walking and biking, and reduce dependence on automobile travel, which is more attractive for
longer commute trips. It would also help to lower vehicle miles traveled and improve air
quality. However, the modeled percentage of intrazonal trips estimated from DRCOG's travel
demand model for the Gates-Cherokee development, a mixed use development planned for a
brownfield site in Denver, was lower than expected. To improve the model results, Cambridge
Systematics developed mixed land use indices to measure land use mix, which could be used to
adjust the intrazonal trips estimated from the gravity trip distribution model.
Based on the analyses described below, the available data did not support the use of an
adjustment factor to intrazonal trips based on any of the mixed use indices studied. It may be
that the lack of correlation between MUI and intrazonal trip rates was due at least in part to the
effects of zone size. In addition, the DRCOG model currently considers only motorized trips,
and a large portion of intrazonal trips are made by walking or bicycling. This might explain the
underestimation of intrazonal trip rates. Therefore, the intrazonal adjustment to the DRCOG
model was not used.
However, to improve the model's sensitivity to changes in land use pattern, DRCOG did alter
its traditional method of model execution in one important respect, balancing its trip generation
model results to attractions rather than to productions. Conventional trip generation models
estimate productions based on trip generation rates and households — in Denver's case, the
number of households as well as income classification and household size. Such models also
estimate trip attractions based on attraction rates and employment—in Denver's case, the model
includes different attraction rates for different area types (CBD, urban, suburban, etc.)
Balancing to productions is often considered the conservative approach as it is generally
believed that household production rates are better known than attraction rates (since typical
travel surveys focus on household travel diaries). However, since the Denver production model
does not include sensitivity to area type, balancing to productions eliminates an important
element of sensitivity to changes in land use pattern. For the model runs conducted for this
project for all scenarios, the trip generation model therefore balanced the trip generation results
to attractions.
Development and Calculation of Mixed Land Use Indices
Cambridge Systematics developed two mixed land uses indices for this project. Each of these
indices requires weights to be computed for each of three land use types: residential, retail, and
C-7
-------
other. Table C.4 summarizes the average number of trip productions and attractions generated
by trip purpose for each of these land use types.
MUIi
A proposed land use mix index was defined by a two-dimensional mixture between industrial
and residential land uses. Suppose that there is a three-dimensional land use mixture:
residential (Xj), retail (X2), and other (X3), where the variables X, represent the percentages of
each land use, weighted by the relative contributions of each land use to trip making. The
weights are computed by each land use type using the trip production and attraction rates in
the existing DRCOG model, as discussed above.
The mixed use ratio (Yij) is computed for each pair of land uses in the same way as in the two-
dimensional mixture:
If Xi=X2=0/ then Yi2=0; Else u
_ ^ 2 ^ ^ 3
If X2=X3=0/then Y23=0; Else 23 ^2+^3
If X1=X3=0/ then Yi3=0; Else 13
Table C.4 Average Productions and Attractions by Land Use Type
Trips Generated per
Retail
Household Employee
Productions (all purposes)
Home based work attractions
Home based non-work
attractions
Non-home based attractions
Total Trips
Weight
8.714(2)
0.035
0.540
0.486
9.775
0.375
1.389
6.477
4.205
12.071
0.464
Non-Retail
Employee (1)
1.389
1.685
1.114
4.188
0.161
Notes:
1. Non-retail employment in the DRCOG model is separated into service and other employment. The
figures in Table 3.1 reflect weighted averages of the contributions of these two employment types.
2. Trip production rate per household weighted over the cross-classifications of income level and
household size used in the DRCOG model
-------
The combined ratio of mixed household, retail, and other land uses can be calculated using
the following equation:
MUIi = (Y12+Y23+Y13) x 2
where:
X] = Ratio of zonal residential land use
X2 = Ratio of zonal retail land use
X3 = Ratio of zonal other industrial land use
¥12 = Mixed use index between residential use and retail industrial use
Y23 = Mixed use index between retail land use and other industrial use
Y13 = Mixed use index between residential use and other industrial use
= Mixed land use ratio 1 for residential, retail industrial and all other use
Assuming residential, retail, and other use are the only three types of land use in a zone,
the proportion of each land use types X; (i=l, 2, 3) has to fall within the following
constraints:
Xi, X2, Xa >=0
Xi+X2+X3 = 1
MUI2
Cambridge Systematics developed an alternative mixed use index. This index is based on
the idea of square deviation about the mean, which is defined as the difference between
the mean and the ratio of an individual land use type. If residential, retail and other land
uses are evenly mixed in a region, the mean ratio of land uses X will be 1/3. The square
deviation of each land use ratio can be calculated as follows:
Z2=(x2-x)2=(x2--)2
Z3=(X3-X)2=(X3-±)2
C-9
-------
The overall mixed land use index is calculated by adjusting the sum of the squared
deviation about the mean:
where:
Xi = Ratio of zonal residential land use
X2 = Ratio of zonal retail land use
Xs = Ratio of zonal other land use
Zj = Mixed use index between residential use and retail use
Z2 = Mixed use index between retail land use and other use
Z3 = Mixed use index between residential use and other use
MUI2 = Mixed land use ratio 2 for residential, retail, and all other use.
The calculation of MUh also assumes residential, retail, and other use as the only three
types of land use in a zone, therefore the proportions land use types X; (i=l, 2, 3) have to
fall within the same constraints as in MUh-
Xi, X2, Xs >=0
Xi+X2+X3 = 1
Range of the MUIs
Both MUIi and MUI2 range between 0 and 1, with 0 meaning the least mixed development
and 1 meaning a completely even mixture. Both MUI calculations give a higher number
for zones that are more mixed-use, as well as those with more trip making. A lower mixed
use index indicates that a dominant land use exists. This can be illustrated through the use
of the following three examples, which represent three different levels of land use mixture
using both of the MUIs:
Case 1: When all three types of land use are evenly mixed (Xi=X2=X3=l/3), we will have
the maximum MUIs:
712 = 723 = Y,,=Y6 and MUI, = 1
Z, = Z2 = Z3 = 0 and MUI2 = 1
Case 2: When there is one and only one type of land use in a zone (for example,
X, = X2 = 0 and X3 = 1, i.e., a zone with only non-retail land uses), we will have the
minimum MUIs:
712 = 723 = 713 = 0 and MUI, = 0
C-10
-------
Zj = Z2 = % and Z3 = % awJ Ml/72 = 0
Case 3: When residential area accounts for V2 of the land, retail and other industrial land
use each takes up 1A of the area (Xi=1/2 and X2=X3=1/4), we will have the following MUIs:
v -Y =V and Y2,= V. and MUL =lj/_ = 0.9167
lz 1J /o Z3 /o ' /lz
Z, = I/ awJ Z2 =Z, = ]/,, awJ ML/72 =0.9375
1 z 3 z
2 , ,,
Jo z 3 7144
Of these three hypothetical cases, Case 1 is the most mixed land use, Case 2 is the least
mixed land uses, and Case 3 is in between. The values of MUIi and MUk reflect the degree
of mixture.
Calculation of Zonal MUIs for Year 2001 and Year 2025
Each version of the mixed land use index was computed for all zones for the 2000 and 2025
scenarios based on land use data provided by DRCOG. The overall results from both
MUIs are very similar. Figures B.I and B.2 illustrate the aggregated MUIs by county for
2000 and 2025, respectively. These figures indicate that all counties in the region are
expected to have a more mixed land use pattern over time. Figures C.6 and C.7,
respectively, show maps of MUIi at the zonal level for 2000 and 2025. Figures C.8 and C.9,
respectively, show maps of MUb at the zonal level for 2000 and 2025.
C-ll
-------
Figure C.4 MUIi by County for Year 2000 and Year 2025
Mixed Land Use Index (1) in Year 2001 and Year 2025
0.60
Adams Arapahoe Boulder Denver Douglas Jefferson Weld N/A All
Counties DYear 2001 BYear 2025 I
Figure C.5 MUI2 by County for Year 2000 and Year 2025
Mixed Land Use Index (2) in Year 2001 and Year 2025
0.70
Adams Arapahoe Boulder Denver Douglas Jefferson Weld N/A All
Counties n Year 2001 Year 2025
C-12
-------
Figure C.6 Distribution of Zonal MUIi for Year 2000
Mixed Land Use Index (1) for Year 2001
I I Counties
MUI1_2001
^10.0-0.2
H 0.2-0.4
0.4-0.6
0.6-0.8
• 0.8-1.0
Figure C.7 Distribution of Zonal MUIi for Year 2025
Mixed Land Use Index (1) for Year 2025
I I Counties
MUI1_2025
| 10.0-0.2
S 0.2-0.4
0.4-0.6
0.6-0.8
• 0.8-1.0
C-13
-------
Figure C.8 Distribution of Zonal MUI2 for Year 2001
Mixed Land Use Index (2) for Year 2001
I I Counties
MUI2_2001
rn P.O. 0.2
a 0.2 - 0.4
0.4-0.6
0.6-0.8
• 0.8 -1
Figure C.9 Distribution of Zonal MUI2 for Year 2025
Mixed Land Use Index (2) for Year 2025
I—I Counties
MUI2_2025
I 10.0-0.2
y 0.2-0.4
0.4-0.6
0.6-0.8
• 0.8-1
C-14
-------
Calculation of Intrazonal Trips from Household Travel Survey and Gravity Model
The next step was to calculate the percentage of intrazonal trips from the household travel
survey and the DRCOG gravity model.
First, the percentage of intrazonal trips for each zone from the household travel survey was
calculated. The household survey is a survey of persons in the households within the region on
the characteristics of the household, characteristics of each person residing in the households
and any out-of-region visitors staying at the households, and the characteristics of travel made
by the person living at the household on the travel survey day. Based on the household travel
survey database, the percentages of intrazonal trips for the three trip purposes in the DRCOG
model—home based work (HBW), home based non-work (HBNW) and non-home based
(NHB) —were calculated. Next, the intrazonal trip percentages based on DRCOG's trip
distribution model were calculated.
Figures C.10 through C.13 compare the county-level intrazonal trip percentages calculated from
the household travel survey with the intrazonal trip percentages estimated from the gravity
model. It is evident from these figures that the gravity models underestimated intrazonal trip
rates for all trip purposes.
Intrazonal Trip Adjustment Models
The calculated MUIs can be used to adjust the intrazonal shares coming out of the gravity
model, so as to overcome the underestimation of intrazonal trip rates. The hypothesis is that
intrazonal trip rates should be higher in areas with higher mixture of land use.
An intrazonal trip adjustment model was estimated for each trip purpose by regressing the
natural log transformation of intrazonal trips from the household survey against the natural log
transformation of intrazonal trips from the gravity model as well as the mixed land use index.
The regression models look like the following:
Ln(intrazonal % from HH survey) = A*Ln(intrazonal % from gravity model) + B*Ln(MUI) + C
where A, B and C are estimated parameters. Adjusting the currently modeled intrazonal
percentages, rather than estimating new percentages based only on the MUI, would allow the
use of other variables that are currently used in estimating intrazonal travel, such as distance to
nearby zones.
The intrazonal trip percentage from the household travel survey was plotted against the
calculated MUIs. As seen in Figures C.14 and C.15, no strong correlation was found between
the percentage of intrazonal trips and either of the mixed use indices. In Figure C.I6, the
observed intrazonal trip percentage was plotted against estimated intrazonal trip percentage
from the gravity model. These correlations seem to be weak as well.
C-15
-------
Figure C.10 Comparison of All Intrazonal Trips
70.00
Intrazonal Trip Percentages (All Trips)
from Gravity Model and the Household Travel Survey
Adams Arapahoe Boulder Denver Douglas Jefferson
Counties
Weld
HH Survey Gravity Model
Figure C.ll Comparison of Home-Based Work Intrazonal Trips
45.00
Intrazonal Trip Percentages (HBW Trips)
from Gravity Model and the Household Travel Survey
Adams Arapahoe Boulder Denver Douglas Jefferson
Counties
Weld
I • HH Survey • Gravity Model
C-16
-------
Figure C.12 Comparison of Home-Based Non-Work Intrazonal Trips
80.00
Intrazonal Trip Percentages (HBNW Trips)
from Gravity Model and the Household Travel Survey
Adams Arapahoe Boulder Denver Douglas Jefferson
Counties
Weld
QHH Survey • Gravity Model
Figure C.13 Comparison of Non Home Based Intrazonal Trips
Intrazonal Trip Percentages (NHB Trips)
from Gravity Model and the Household Travel Survey
80.00
Adams Arapahoe Boulder Denver Douglas Jefferson
Counties
Weld
HH Su rvey Gravity Model
C-17
-------
Figure C.14 Plot of Observed Intrazonal Trip Percentage and Calculated MUIi
80
1
OT
ID
co 60
H
T3
O
o 40
_c
E
o
•^ 20
o
N
ro
0.0
1.0
1.2
Mixed land use index 1, year 2001
C-18
-------
Figure C.15 Plot of Observed Intrazonal Trip Percentage and Calculated
80
>
2 60
-I—I
2
o
o
.c
E
o
.fc
o
N
CD
40
20
0.0
.4
.6
1.0
Figure C.16 Plots of Observed and Estimated Intrazonal Trip Percentages
CO
-fc
80
60
O 40
E
o
J=
CO
c
o
N
CO
D_
D
CD D
-10 0 10 20 30 40
Intrazonal trip % from gravity model, year 2001
50
C-19
-------
Revision of Mixed Land Use Indices
It was hypothesized that the lack of correlation between MUI and intrazonal trip rates was due
at least in part to the effects of zone size. The original calculations of MUIi and MUb do not
account for zone size. More rural zones are generally large in area and tend to have mixed uses,
even when those uses could be miles apart. Therefore, the calculation of mixed land use indices
was revised to take zone size into account. The idea was that larger zones provided more
opportunities for intrazonal travel. For example, in both of the original measures, a zone with
10 households and 10 retail employees would have the same MUI value as a zone with 100
households and 100 retail employees. But the larger zone could in reality provide more
opportunities for intrazonal travel.
To account for zone size, the MUI calculations were revised by using the number of households,
retail employment, and other employment instead of percentages of these land use types. The
question is, would zones that qualify as mixed use more because of their size than because they
represent mixed use development be likely to have higher percentages of intrazonal trips than
non-mixed use zones. This may be the case, in part because large zones would be more likely to
have intrazonal trips. A trip to the supermarket may be a few miles long but still be intrazonal.
Model Results Discussion and Recommendations
Unlike the CTPS model in Boston, the DRCOG model currently considers only motorized trips,
and a large portion of intrazonal trips are made by walking or bicycling. This might explain the
underestimation of intrazonal trip rates. Regardless of the reason, it is clear that the available
data did not support the use of an adjustment factor to intrazonal trips based on any of the
mixed use indices studied. Therefore, the intrazonal adjustment to the DRCOG model was not
used.
Trip Generation Revision
The second model enhancement analyzed for Denver is to adjust non-motorized trip rates. The
idea behind this enhancement is to consider the hypothesis that pedestrian/bicycle friendly
areas should have more non-motorized trips and, therefore, fewer motorized trips. Since the
DRCOG model considers only motorized trips, the implication is that (motorized) trip
generation rates should be lower in pedestrian or bicycle friendly zones. In model application,
these lower rates are applied to zones identified as pedestrian/bicycle friendly.
To implement this model enhancement, Boulder County was used as a test area. The trip rates
from the household survey in Boulder County for households in pedestrian/bicycle friendly
zones were compared to the rates for Boulder County households in other zones. Because trip
rates for attractions and non-home based productions in the DRCOG model already were based
on area type, only the home based work and home based non-work production rates were
adjusted.
Trip production rates in the DRCOG model are cross-classified by income level and household
size. Separate adjustments, however, could not be computed for each cell in the cross-
classification due to insufficient data in the household survey data set.
Pedestrian or Bicycle Friendly Zones in Boulder County
DRCOG staff identified zones in Boulder County that are considered pedestrian or bicycle
friendly. Three levels of pedestrian or bicycle friendliness were identified as in the following:
C-20
-------
• Level I, the most pedestrian-friendly zones (eight zones)
• Level 2, zones that are somewhat pedestrian-friendly (nine zones)
• Level 3, all other zones in Boulder County
To validate the assumption that these three levels of zones have different levels of
pedestrian/bicycle friendliness, the home based work and home based non-work trip records
person trip records in the household travel survey that correspond to the Boulder County zones
were identified. Households with students at Colorado University were excluded. From these
records, the average motorized and non-motorized trip rates by purpose for each zone type
were computed.
Table C.5 shows the ratios of non-motorized trips by purpose for zones with different levels of
biking/pedestrian friendliness. This table indicates that non-motorized trips in the pedestrian
or bicycle friendly zones are higher than average for all trip purposes. For the Level 1 zones,
non-motorized trips account for 40 percent of all home-based work trips, 44 percent of home-
based non-work trips, and 21 percent of non-home based trips. In contrast, the average regional
non-motorized trip ratio is only 6 percent for all home-based work trips, 12 percent for home-
based non-work trips, and 13 percent for non-home based trips.
Trip Adjustment Factor Calculation
The formula for calculating the trip rate adjustment factor for each trip purpose is:
MOTOR „ X MOTOR
(MOTOR v + NONMOTOR v ) £ (MOTOR v + NONMOTOR
where:
i - Zone types (i = 1, 2, or 3)
j - Trip purposes (j=l for HBW, j=2 for HBNW and j=3 for NHB)
Ay - Motorized trip adjustment factors for zone type i and trip purpose j
MOTOR;] - Motorized trip number for zone type i and trip purpose j
NONMOTORy - Non-Motorized trip number for zone type i and trip purpose j
- Total Motorized trips for all zones with trip purpose j
Y (MOTOR + NONMOTOR ) " Total trips (motorized and non-motorized) for all
zones with trip purpose j
This formula represents the ratio of the motorized mode share for the zone relative to the
motorized mode share for all zones. Table C.6 presents the calculated adjustment factors
for all three types of zones by trip purpose.
It is interesting to note the counterintuitive result for home based non-work trips, where
the adjustment for Level 2 zones is more significant than for Level 1 zones. To address this
C-21
-------
concern, factors were computed for a two-tiered zone system, where Levels 1 and 2 were
combined. The resultant adjustment factors are presented in Table C.7.
Because of the inconsistent result for home based non-work trips, DRCOG and Cambridge
Systematics decided to use the two-tiered zone classification system and the adjustment
factors shown in Table C.7. The DRCOG trip generation program was revised to
incorporate this revision.
Table C.5 Percentage of Trips in Boulder County That Are Non-Motorized
Zone Type
Level 1
Level 2
Level 3
All Levels
HBW
40%
21%
4%
6%
HBNW
44%
34%
10%
12%
NHB
21%
28%
12%
13%
Table C.6 Calculated Adjustment Factors for the Three-Tiered Zone System
Zone Type Home Based Work Home Based Non-Work
Level 1
Level 2
Level 3
0.653
0.835
1.016
0.789
0.726
1.012
Table C.7 Calculated Adjustment Factors for the Two-Tiered Zone System
Zone Type Home Based Work Home Based Non-Work
Levels 1/2 0.779 0.739
Level 3 1.016 1.012
Emissions Analysis Methods
The Denver emissions analysis was performed by the Colorado Department of Public Health,
which undertakes emissions modeling for the Denver Regional Council of Governments
(DRCOG) and other transportation agencies in Colorado. The analysis was performed using
MOBILE6, with updated Denver-specific inputs for vehicle registration distribution (based on
year 2000 State of Colorado data), VMT mix by road type and time of day, I/M program, and
fuel characteristics. Average speeds by facility type were used for the ten time periods modeled
by DRCOG. Start-based emissions were not modeled separately from VMT-based emissions.
C-22
-------
Appendix D - Benefits of Using MOBILE6
EPA's MOBILE6 emissions factor model incorporates numerous enhancements to the previous
MOBILES model, many of which are helpful in examining the emission impacts of a regional
policy of promoting brownfield and infill development. Table A.I summarizes the ability of
MOBILE6 and MOBILES to capture the travel effects typically associated with brownfield and
infill development.
MOBILE6 produces different emission factors due to differences in emissions modeling
assumptions. A number of states and urban areas have found MOBILE6 to produce higher
emission factors in the short term and lower emission factors in the long term compared to
MOBILES. As a result, the absolute emission benefits of a buildout (long-term) analysis could
be overstated by the use of MOBILES. However, it is not likely that percentage differences
between scenarios would differ significantly.
In summary, MOBILE6 provides the following specific benefits over MOBILES for analyzing
the emissions impacts of brownfield and infill developments:
• Allows the use of start-based emission factors as opposed to those based on vehicle-
miles traveled (VMT). This is important if infill developments reduce vehicle trips in
different proportion to VMT (i.e., if vehicle trip-lengths are shorter but there are just as
many vehicle trips or, conversely, if vehicle trip lengths remain the same but there are
fewer vehicle trips.) MOBILES does not provide separate emission factors for trip ends
versus vehicle-miles of travel. As a result, the use of MOBILES could potentially
underestimate the benefits of strategies that reduce vehicle trips that are shorter than
average (shorter trips would have higher per-mile emissions, because of the contribution
of start emissions).
• MOBILE6 produces facility-specific speed-based emission factors. MOBILE6 also
contains updated/improved speed correction factors (SCFs) that vary by facility type.
Thus, MOBILE6 provides more reliable estimates of the effects of changes in average
vehicle speeds on emissions and also allows shifts in traffic among road types (freeways,
arterials, local) to be assessed. These enhancements could lead to different estimates of
the benefits of strategies that affect vehicle speeds and also of strategies that shift trips
from one facility type to another. The impact of changes in speeds is likely to be smaller
when estimated using MOBILE6 than for MOBILES. Without looking at results in detail,
however, it is difficult to say whether the speed and facility type changes alone lead to
increases or decreases in emissions between scenarios.
• MOBILE6 allows changes in the distribution of trip lengths to be assessed (although this
would require additional processing of travel analysis outputs).
D-l
-------
Table D.I Benefits of Using MOBILE6 versus MOBILES for Analyzing Infill/
Brownfields Vehicle Emissions Impacts
Effect on Travel Patterns
Do common travel analysis Can MOBILES
methods give us this Measure?
information?
Can MOBILE6
Measure?
1. Shorter trip lengths due to:
a. Regional context/
accessibility
b. Local street connectivity
2. Fewer vehicle trips due to:
a. Mixed-use development
and pedestrian access
b. Transit accessibility
(regional)
3. Different traffic/driving
patterns
a. Slower average trip speeds Yes (within limitations of
due to urban setting speed output of 4-step
models)
b. Different mix of travel by Yes (for freeways &
roadway type (e.g., more arterials - but some
local, less freeway) limitations for local roads)
Yes (interzonal trips are
modeled)
No (unless intrazonal trip
lengths are adjusted)
Possibly (adjustments such
as pedestrian environment
factors must be applied)
Yes
Yes (VMT-based Yes/improved
emission factors) (effects of changes in
distribution of trip
lengths)
No
Yes (speed
correction
factors)
No
Yes (Start-based
emission factors)
Yes/improved
(updated speed
correction factors)
Yes (speed
correction factors by
roadway type)
c. Lower acceleration and No
deceleration rates due to
lower speeds, urban setting
No (except as
embodied in
SCF)
No (except as
embodied in SCF by
roadway type)
The benefits of applying MOBILE6 instead of MOBILES depend on having reliable data on
travel pattern impacts, including effects of differences in development patterns and urban
design on vehicle trip-making, vehicle trip lengths, and vehicle travel speeds by facility. The
ability of existing analysis methods to assess these parameters varies.
D-2
-------
Brownfield and infill developments also may have travel impacts that are not easily and
routinely analyzed by common travel demand analysis methods, such as:
• Changes in vehicle activity patterns (e.g., fewer peak-hour vehicle trips, fewer "cold-
start" trips, different soak times.)
• Differences in commercial vehicle trip rates or trip patterns.
• Changing patterns of vehicle ownership and use (e.g., fewer miles per year per car).
If data are available on these travel impacts, MOBILE6 can be used to assess the resulting
emissions impacts (generally, more effectively than MOBILES), but these data are not available
in the travel modeling done for any of the three participating partner communities.
D-3
------- |