Transportation safety has been recognized as a public health issue worldwide,
consequently, transportation researchers and practitioners have been attempting
to provide adequate safety performance for the various transportation components
and facilities to all road users given the usually scarce resources available. Safety
engineers have been trying to make decisions affecting safety based on the knowledge
extracted from different types of statistical models and/or observational before-after
analysis. It is generally recognized that this type of factual knowledge is not easily
obtained either statistically or empirically. Despite the intuitive link between road
safety and observed crashes, a good understanding of the sequence of events prior
to the crash can provide a more rational basis for the development of engineering
countermeasures.
The development of more comprehensive mechanistic models for safety assessment
is heavily dependent on detailed vehicle tracking data that is not readily
available. The potential of microscopic simulation in traffic safety and traffic conflict
analysis has gained increasing interest mostly due to recent developments in
human behaviour modelling and real-time vehicle data acquisition.
In this thesis, we present a systematic investigation of the use of existing behavioural
microscopic simulation models in short-term road safety studies. Initially,
a microscopic framework is introduced to identify potentially unsafe vehicle interactions
for different vehicle movements based on three types of traffic behaviour
protocols: car-following, lane change and gap acceptance. This microscopic model
for safety assessment applies a safety performance measure based on pairwise comparisons
of spacing and speed differential between adjacent vehicles and individual
braking power in real-time. A calibration/validation procedure using factorial analysis
is presented to select best model input parameters for this safety performance
measure by using high resolution vehicle tracking data. The ability of the proposed
safety performance measure to reflect real-life observed high-risk vehicular
interactions is explored in three intuitive tests using observed crash data. Finally,
the usefulness of the model is illustrated through its application to investigate the
safety implications of two different geometric and operational traffic strategies.
The overall results indicate that, notwithstanding the fact that actual behavioural
microscopic algorithms have not been developed strictly to model crashes, they are
able to replicate several factors directly related to high risk situations that could
lead to crashes with reasonable accuracy. With the existing upward trend in computing
power, modelling techniques and increasing availability of detailed vehicle
tracking data, it is likely that safety studies will be carried out using a more mechanistic
and inclusive approach based on disruptive driving behaviour rather than
ultimate unpredictable and heavily restrictive crash events.
Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:OWTU.10012/4111 |
Date | January 2008 |
Creators | Cunto, Flávio |
Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
Language | English |
Detected Language | English |
Type | Thesis or Dissertation |
Page generated in 6.5096 seconds