6 Disequilibrium Network Design: A New
Paradigm for Transportation Planning and
Control
Terry L. Friesz
Departments of Systems Engineering
and Operations Research & Engineering
George Mason University
Fairfax, VA 22030, USA
6.1 Introduction
Samir Shah
PB Farradyne Inc
Rockville, MD 20852
USA
David Bernstein
Dept. of Civil Engineering
and Operations Research
Princeton University
Princeton, NJ 08544, USA
In a very broad sense, the network design problem (NDP) is a topic that has captured
the attention of many researchers. This is mainly due to the immense importance
of strategic capital investment decisions involving transportation infrastructure. For
the purposes of this paper we limit the scope of the network design problem to
highway systems. Design related decisions relevant to highways include a rich and
wide variety of strategic (e.g., new right of way), tactical (one way street assignment,
HOV assignment) and operational (traffic signalization, ramp metering) decisions
that typically arise in transportation planning.
The equilibrium network design problem is to find an optimal network design in
terms of additional facilities or capacity enhancements, when the network flow
pattern is constrained to be an equilibrium. The much reported occurrence of
Braess' paradox (Murchland, 1970) requires that design models have Wardropian
user equilibrium (Wardrop, 1952) constraints. The transportation research literature
includes many such equilibrium design models wherein equilibrium constraints are
generally articulated as an equivalent optimization problem (Abdulaal and LeBlanc,
1979; LeBlanc,1975) or as an equivalent variational inequality problem (Friesz et
aI., 1990; 1992; 1993b; Marcotte, 1986). Unfortunately such equilibrium network
design models presuppose a static environment and completely ignore the impacts
of potential disequilibria which can arise due to perturbations in the capacity of the
network infrastructure. Such a static perspective may lead to the occurrence of a
temporal version of Braess' Paradox. The "temporal Braess' Paradox" does not
yet enjoy a standard definition, but in the context of the present discussion can be
viewed as occurring when a capacity altering action lowers (or leaves unchanged)
overall delay in the present and near future but increases overall delay at some more
distant time. Such outcomes are possible when the present value of disequilibrium
impacts is substantially negative, as can occur when the immediate disequilibrium
response is a sharp congestion increase or when the disequilibrium response is a
mild congestion increase of relatively long duration. An example is provided by a
highway construction project intended to enhance capacity but which produces traffic
congestion for weeks or months prior to its completion; in this case, the present value
of construction impacts may be sufficiently negative to eradicate all positive benefits
in the post-construction period. This is particularly true in standard cost benefit kind
of analyses, wherein near term benefits are more heavily discounted than long term
L. Lundqvist et al. (eds.), Network Infrastructure and the Urban Environment
© Springer-Verlag Berlin · Heidelberg 1998