Ph.D. Thesis Defense Announcement
Macroscopic Urban Network Dynamics: Estimation and Applications
By:
Rafegh Aghamohammadi
Advisor:
Dr. Jorge A. Laval (CEE)
Committee Members:
Dr. Srinivas Peeta (CEE), Dr. Patricia Mokhtarian (CEE),
Dr. David Goldsman (ISyE), Dr. Guanghui Lan (ISyE)
Date & Time:
Friday, December 3rd, 2021, at 11 AM (EST)
Location:
In-person: Sustainable Education Building (SEB) 122
Virtual: https://PhDdefense.rafegh.me
During the past decade there has been significant research efforts in developing traffic control and management
methods based on an aggregated representation of traffic networks. In fact, the traditional link-level network
representation imposes prohibitive computational costs for typical large-scale urban networks. Thankfully, it has
been observed that at a macroscopic level, the relationship between any pair of network-average traffic variables
can be described by simple functions called macroscopic fundamental diagrams (MFD). However, current MFD
estimation methods were mainly conceived for individual arterial corridors and their application to urban networks
has not been validated using extensive empirical data.
This dissertation fills this gap by extending current MFD estimation methods to large-scale real-life networks,
while using empirical data from 41 cities around the world for calibration and validation. This dissertation further
investigates the efficient application of MFD in travelers' route choice using the dynamic traffic assignment (DTA)
methods and sets forth the discrete- and continuum-space DTA approaches are intrinsically similar and can be
seen as equivalents on different aggregation levels, although they previously seemed to be the two extreme ends
of the macroscopic DTA spectrum. A novel continuum-space DTA modeling framework consistent with the MFD
theory and assumptions has been developed and a semi-Lagrangian solution method has been proposed by
splitting up the network into smaller zones, which can be implemented for minimizing either the travel times of
individual users or the total travel time of all users in the network. Finally, the potentiality of implementing the
MFD in microscopic vehicular emissions estimation models has been explored.
The major findings of this dissertation are as follows. The empirical MFD validation results identify the most
important challenges in both analytical and empirical MFD estimation approaches as: (i) the distribution of loop
detectors within the links, (ii) the distribution of loop detectors across the network, and (iii) the treatment of
unsignalized intersections and their impact on the block length. The numerical experiment results using the
proposed DTA framework indicate that partitioning the network into a finer grid of zones can yield more accurate
results with respect to the approximated analytical solutions without significant loss of efficiency and demonstrate
the potential of application of this framework for real-life networks with arbitrary network and zone shapes. The
comparison between the results and runtimes of the emissions estimations conducted in 4 different aggregation
levels: lane, link, corridor, and network, reveals that the efficiency can be significantly improved by utilizing more
aggregated network representation under some considerations. This will make the MFD a powerful tool for real-
time emissions estimation and control.
http://PhDdefense.rafegh.me