Strategies, models, and algorithms facilitating such models are explored
to provide transportation network managers and planners with more
flexibility under uncertainty. Network design problems with
non-stationary stochastic OD demand are formulated as real option
investment problems and dynamic programming solution methodologies are
used to obtain the value of flexibility to defer and re-design a
network. The design premium is shown to reflect the opportunity cost of
committing to a “preferred alternative” in transportation planning.
Both network option and link option design problems are proposed with
solution algorithms and tested on the classical Sioux Falls, SD network.
Results indicate that allowing individual links to be deferred can
have significant option value.
A resource relocation model using non-stationary stochastic variables as
chance constraints is proposed. The model is applied to air tanker
relocation for initial attack of wildfires in California, and results
show that the flexibility to switch locations with non-stationary
stochastic variables providing 3-day or 7-day forecasts is more
cost-effective than relocations without forecasting.
Due to the computational costs of these more complex network models, a
faster converging heuristic based on radial basis functions is evaluated
for continuous network design problems for the Anaheim, CA network with
a 31-dimensional decision variable. The algorithm is further modified
and then proven to converge for multi-objective problems. Compared to
other popular multi-objective solution algorithms in the literature such
as the genetic algorithm, the proposed multi-objective radial basis
function algorithm is shown to be most effective.
The algorithm is applied to a flexible robust toll pricing problem,
where toll pricing is proposed as a strategy to manage network
robustness over multiple regimes of link capacity uncertainty. A link
degradation simulation model is proposed that uses multivariate
Bernoulli random variables to simulate correlated link failures. The
solution to a multi-objective mean-variance toll pricing problem is
obtained for the Sioux Falls network under low and high probability
seasons, showing that the flexibility to adapt the Pareto set of toll
solutions to changes in regime – e.g. hurricane seasons, security threat
levels, etc – can increase value in terms of an epsilon indicator.
With soaring oil prices and growing concerns for global warming, there
is increasing interest in the environmental performance of
transportation systems. This dissertation contributes to this growing
literature through three independent yet related projects essays that
deal with transportation technology, infrastructure, and policy.
My first essay analyze the increasing interest for hybrid cars by
Californians based on a statewide phone survey conducted in July of 2004
by Public Policy Institute of California (PPIC) using discrete choice
models. Results suggest that the possibility for single drivers to use
hybrid vehicles in HOV lanes is more important than short term concerns
for air pollution, support for energy efficiency policies, long term
concerns for global warming, education, and income. This suggests that
programs designed to improve the environmental performance of individual
vehicles need to rely on tangible benefits for drivers; to make a
difference, they cannot rely on environmental beliefs alone.
The second essay is concerned with assessments of Travel Demand
management (TDM) policies, which have been used to deal with congestion,
air pollution, and now global warming. I compare two TDM programs: Rule
2202 (The on-road motor vehicle mitigation options in southern
California) and the Commute Trip Reduction Program (CTR) in Washington
State. My results reveal that after 2002, the impacts of Rule 2202 are
mixed. Commuters’ modal choices are affected by worksite
characteristics but only two (out of six) basic strategies effect the
change in average vehicle ridership (AVR). Moreover, the level of
subsidies appears to play an important role in commuting behavior. In
Washington State, location has an impact on AVR and combinations of
location and employee duties influence the single occupant vehicle
index. Details of the CTR and its relative success suggest that there
is room for improving Rule 2202 to make it friendlier to businesses and
more effective.
Finally, I examine the health impacts of NOx (nitrogen oxides) and PM
(particulate matter) generated by trains moving freight through the
Alameda Corridor to and from the Ports of Los Angeles and Long Beach.
After estimating baseline emissions for 2005, I examine two scenarios:
in the first one, I assume that all long-haul and switching locomotives
are upgraded to Tier 2 (from Tier 1); in the second scenario, all Tier 2
locomotives operating in the study area are replaced with cleaner, Tier
3 locomotives. I find that mortality from PM exposure accounts for the
largest component of health impacts, with 2005 annual costs from excess
mortality in excess of $40 million. A shift to Tier 2 locomotives would
save approximately half of these costs while the benefits of shifting
from Tier 2 to Tier 3 locomotives would be much smaller. To my
knowledge, this is the first comprehensive assessment of the health
impacts of freight train transportation in a busy freight corridor.
Characteristics of the built environment, such as the mixture of land uses, transportation infrastructure, and neighborhood design, have often been associated with reduced automobile use and increased walking and transit use. However, a significant gap remains in our understanding of travel behavior, especially with respect with social environmental and attitudinal factors influencing travel, such as crime rates and the perceptions of walking. This dissertation, comprised of four empirical essays, explores the complex relationships between the built and social environment and neighborhood travel by focusing on non-work travel for individuals sampled from eight communities in the South Bay area of Los Angeles County.
In the first essay, I examine claims made by proponents of New Urbanism that traditional neighborhood designs promote walking and discourage driving by comparing automobile and walking trip rates for mixed-use centers and auto-oriented corridors. The results showed no discernable differences in individual driving trips between these two types of neighborhoods while more walking trips were reported in mixed-use centers. Therefore, the results both support and challenge New Urbanist claims.
The second essay examines the interactions between race/ethnicity, demographic change, and travel behavior by comparing driving and walking trips across racial and ethnic groups. The results showed that African-Americans took fewer driving trips and Asians walked less compared to non-Hispanic whites, and that Hispanics who walk are more sensitive to demographic changes in their neighborhood than other groups.
The third essay focuses on crime and perceptions of safety and how they impact walking behavior. After taking sociodemographic and built environment factors into account, violent crime rates had a strong deterrent effect on walking across race, income, and gender groups, while perceptions of neighborhood safety varied.
In the fourth essay, I focus on whether the built environment encourages walking above and beyond individuals’ attitudes toward walking. By comparing individuals with positive attitudes toward walking with those with neutral or negative attitudes, the results showed that individuals with positive attitudes were more responsive to built environment characteristics than those held negative attitudes. These findings suggest differences in walking behavior are more strongly shaped by personal attitudes than the built environment.
Use of advanced traffic control systems ranks as one of the most
cost-effective actions for urban transportation improvements to mitigate
total delay and alleviate fuel consumption and air pollution.
Nonetheless, Adaptive Signal System, the most advanced type of traffic
control designed for real-time traffic responsive operations, is not
widely accepted in field implementation. Benefits of such systems are
not fully realized yet, mainly because of the large cost for installment
and maintenance of required sensor systems for traffic forecast.
Moreover, even with the sensor systems, the performance still suffers
due to inaccurate prediction caused by the limitation of data sources
and deficiencies in the control algorithms.
Based on these observations, this study developed the applications of
emerging data sources in traffic control system. Traffic parameters are
collected under the new traffic information system such as a Persistent
Traffic Cookies (PTC) system conceptually proposed at UC Irvine using
wireless communication between a vehicle and a roadside hardware. With
the preliminary study results under the system, this study develops
traffic control schemes with the traffic forecast resulting from the PTC
system. Initially, general methods are presented to generate required
input, that is path-based traffic variables such as the turning flows
and travel time from PTC data. The inputs were implemented in two
different traffic control schemes; subnetwork definition for
area-control and signal optimization scheme in network-level. The
relevant spatial boundary for area-control is determined by a systematic
approach on the basis of traffic dynamics estimated by the PTC data.
Basically, the approach is to group multiple interconnected
intersections with strong control dependencies on each other, which can
be measured by the path flow among the intersections. Another
application is a signal optimization scheme at the network-level under
the assumption of fully decentralized control embedded with indirect
signal coordination consideration. Local optimization was accomplished
by a Dynamic Programming approach incorporating with a modified Rolling
Horizon Scheme and network-wide coordination was indirectly achieved by
iterative approach with repeated local optimizations.
For an evaluation of proposed control scheme, a simulation study was
presented using Irvine Triangular Network constructed in microscopic
simulation software. Results show that the proposed scheme is capable
of reducing total delays in a network, in comparison to Actuated Signal
Control already installed in the study network. It is also shown that
the scheme that incorporates certain modified rolling horizon methods
performs better than that with a more conventional rolling horizon method.
A spectrum of traffic engineering and modern transportation planning problems requires the knowledge of the underlying trip pattern, commonly represented by dynamic Origin-Destination (OD) trip tables. In view of the fact that direct survey of trip pattern is technically problematic and economically infeasible, there have been a great number of methods proposed in the literature for updating the existing OD tables from traffic counts and/or other data sources. Unfortunately, there remain several common theoretical and practical aspects which impact the estimation accuracy and limit the use of these methods from most real-world applications. This dissertation itemizes and examines these critical issues. Then, the dissertation presents the developments, evaluations, and applications of two new frameworks intended to be used with the current and near-future data, respectively.
The first framework offers a systematic and practical procedure for preparing dynamic demand inputs for microscopic traffic simulation under planning applications with an estimation module based solely on traffic counts. Under this framework, the traditional planning model is augmented with a filter traffic simulation step, which captures important spatial-temporal characteristics of route and traffic patterns within a large surrounding network, to improve the flow estimates entering and leaving the final microscopic simulation network. A new bounded dynamic OD estimation model and a solution algorithm for solving a large problem are also proposed.
The second framework utilizes additional information from small probe samples collected over multiple days. There are two steps under this framework. The first step includes a suite of empirical and hierarchical Bayesian models used in estimating time-dependent travel time distributions, destination fractions, and route fractions from probe data. These models provide multi-level posterior parameters and tend to moderate extreme estimates toward the overall mean with the magnitude depending on their precision, thus overcoming several problems due to non-uniform (over time and space) small sampling rates. The second step involves a construction of initial OD tables, an estimation of route-link fractions via a Monte Carlo simulation, and an updating procedure using a new dynamic OD estimation formulation which can also take into account the stochastic properties of the assignment matrix.
Uncertainties in transportation capacity and cost pose a significant
challenge for both shippers and carriers in the trucking industry. In
the practice of adopting lean and demand-responsive logistics systems,
orders are required to be delivered rapidly, accurately and reliably,
even under demand uncertainty. These tougher demands on the industry
motivate the need to introduce new instruments to manage transportation
service contracts. One way to hedge these uncertainties is to use
concepts from the theory of Real Options to craft derivative contracts,
which we call truckload options in this dissertation. In its simplest
form, a truckload call (put) option gives its holder the right to buy
(sell) truckload services on a specific route, at a predetermined price
on a predetermined date. The holder decides if a truckload option should
be exercised depending on information available when the option expires.
Truckload options are not yet available, however, so the purpose of
this dissertation is to develop a truckload options pricing model and to
show the usefulness of truckload options to both shippers and carriers.
Since the price of a truckload option depends on the spot price of a
truckload, we first model the dynamics of spot rates using a common
stochastic process. Unlike financial markets where high frequency data
are available, spot prices for trucking services are not public and we
can only observe some monthly statistics. This complicates slightly the
estimation of necessary parameters, which we obtain via two independent
methods (variogram analysis and maximum likelihood), before developing a
truckload options pricing formula. A numerical illustration based on
real data shows that truckload options would be quite valuable to the
trucking industry.
This dissertation develops a method to create value through more
flexible procurement contracts, which could benefit the trucking
industry as a whole – particularly in an uncertain business environment.
Truckload rate and truckload options price are solidly investigated
and modeled. In addition, parameter estimation for a continuous
stochastic model is practically explored using discrete statistics.
Finally, numerical trading examples are illustrated and a picture of
truckload option trading becoming reality presented. The complicated
results indicate that truckload options have the potential of
stimulating the entire trucking and logistics industries.
Because of the growing importance of hub-and-spoke operations in the
trucking industry, crossdocking has become an important and effective
tool to transfer freight. Companies like Wal-Mart, Costco and Home Depot
are using this kind of facility in their logistics operations.
Efficiently operating crossdocks, thereby reducing unnecessary waiting
and staging congestion for freight and workers are important issues for
managers.
To deal with the above issues, this dissertation uses real-time
information about the contents of inbound and outbound pallets and the
locations of pallets to schedule unloading for waiting trailers and
assign destinations for pallets. We show how to incorporate the
information of waiting freight in trailers to benefit trailer
scheduling; we also present how to use the information on freight
staging to mitigate congestion. Two dynamic trailer scheduling and four
alternate destination strategies are proposed accordingly and compared
with baseline scenarios in this research.
Our simulation results suggest that:
1. Our strategies are effective. The two time-based trailer scheduling
algorithms can save cycle times as high as 64%, 57% and 30% in the
4-to-4, 4-to-8 and 8-to-8 crossdock scenarios, respectively; the four
alternate destination strategies can save cycle times as high as 34% in
the 8-to-8 staging crossdock scenarios. In addition, these strategies
can raise throughputs for crossdocks. These effects should result in a
noticeable influence on supply chain networks, including shorter
transportation lead-time, more reliable on-time deliveries and less
inventory costs.
2. In our alternate destination strategies, even if a destination change
results in extra time for value-added services for freight, the
strategies are still worthy to adopt.
3. The combination models of our trailer scheduling algorithms and
alternate destination strategies work better than solely implementing an
alternate destination strategy when trailer arrivals are dense.
4. A higher flexibility on choosing alternate destinations can bring
higher performance for crossdocks.
Efficient freight transportation is an essential for a strong economic
system. A rapid growth of freight demand, however, lessens the
efficiency of provided infrastructure. In order to alleviate this
problem effectively, evaluation studies have to be performed in order to
invest the limited budget for the best of social benefits. In addition
to many difficulties on making a decision for each project investment,
it is made harder by the complimentary and substitution effects that
happen when considering transportation project together. Current
practices, however, limit number of project combinations in order to
avoid numerous tests. The best project combination may have never been
realized.
This dissertation proposes network design models which can automatically
create project combinations and searching for the best. The network
design models have been studied for the passenger movements and focus on
highway expansions. In this dissertation, the focus is shifted to the
freight movements which involve multimodal transportation improvements.
Our freight network design model is developed based on the bi-level
optimization model. The development then involves two components. The
first task is to set the freight investment problems within the bi-level
format. This includes finding a suitable freight flow prediction model
which can work well with the bi-level model. The second task is to
provide a solution algorithm to solve the problem.
The dissertation sets the framework of the freight flow network design
model, identifies expecting model issues, and provides alternatives that
alleviate them. Through a series of developments, the final model uses
the shipper-carrier freight equilibrium model to represent freight
behaviors. Capacity constraints are used as a mean to emphasize limited
services since the reliability issues, an important factor for freight
movements, cannot be captured by steady state traffic assignment. A case
study is implemented to allocate a budget for improvements on the
California highway network. The transportation modes are selected by the
shipper model which can be trucks, rails, or the multimodal
transportation. The results shown that the proposed network design model
provided better solutions compared with traditional ranking methods. The
solution algorithm can manage the problem with reasonable project
alternatives. However, the computation expense increases rapidly with
increasing number of project alternatives.
Commercial vehicles typically represent a small fraction of vehicular
traffic on most roadways. However, their influence on the economy,
environment, traffic performance, infrastructure, and safety are much
more significant than their diminutive numerical presence suggests.
This dissertation describes the development and prototype implementation
of a new high-fidelity inductive loop sensor and a ground-breaking
commercial vehicle classification system based on the vehicle inductive
signatures obtained from this sensor technology. This new sensor
technology is relatively easy to install and has the potential to yield
reliable and highly detailed vehicle inductive signatures for advanced
traffic surveillance applications.
The Speed PRofile INterpolation Temporal-Spatial (SPRINTS)
transformation model developed in this dissertation improves vehicle
signature data quality under adverse traffic conditions where
acceleration and deceleration effects can distort inductive vehicle
signatures. The axle classification model enables commercial vehicles
to be classified accurately by their axle configuration. The body
classification models reveal the function and unique impacts of the
drive and trailer units of each commercial vehicle.
Together, the results reveal the significant potential of this inductive
sensor technology in providing a more comprehensive commercial vehicle
data profile based on a unique ability to extract both axle
configuration information as well as high fidelity undercarriage
profiles within a single sensor technology to provide richer insight on
commercial vehicle travel statistics.
The activity-based travel demand model recognizes that travel is derived
from the demand for activity participation distributed in space. The
focus on intra-household interactions and linkages between people’s
behavior and social and physical environment has been identified as
emerging features of the activity-based approach that would be important
to travel behavior research. The dissertation is dedicated to an
in-depth exploration of the within-household interactions by theoretical
specification and empirical development of the household activity time
allocation models based on a utility maximization framework with the
household as the unit of analysis. Furthermore, the dissertation also
aims to propose a model of the household activity scheduling process
primarily focusing on task allocation mechanisms on the basis of the
human agents adjusting themselves to the built social and physical
environment.
Development of the activity time allocation model in this dissertation
includes two types of structural time allocation models. First, the
collective models based on two assumptions that household heads have
their own utility functions and that decisions by them reach
Pareto-efficient outcomes are introduced to develop intra-household
activity time allocation models for leisure demand and housework
activity. Secondly, intra-household time allocation to housework
activity is further examined through the estimation of time allocation
to the different types of activities by the different types of household
members along with extensive exploration of various theories and
identification of related interactions.
This dissertation proposes a household activity scheduling process a
model design based on a weekly pattern system, which is expected to keep
various advantages compared to a deterministic daily model system. Along
with learning and adaptation procedures, the human being as a learning
agent is designed to prepare strategic plans of behavior to achieve
individual goals through interactive environments, and operationalize
those plans via activity execution requiring the participation of other
agents. At the household level, the household and its members as
decision agents are also designed to optimize the allocation of the
available household labor resource under the presence of the
uncertainties of the physical and social environments. After describing
the mathematical framework and solution procedure, a simulation
experiment is conducted within a hypothetical environment to demonstrate
how the proposed model works.