Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

KSCE Journal of Civil Engineering (2016) 20(3):1084-1098 Sustainable Urban Transportation System

Copyright ⓒ2016 Korean Society of Civil Engineers


DOI 10.1007/s12205-016-1729-1 pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

A Review of Sustainable Network Design for Road Networks


Xiangdong Xu*, Anthony Chen**, and Chao Yang***
Received September 8, 2015/Accepted January 7, 2016

··································································································································································································································

Abstract

The Network Design Problem (NDP) is a strategical decision-making problem in planning, designing, and managing road
networks with the aim to make efficient use of limited resources for optimizing the road network performance. Sustainability
development is a major concern of various social-economic systems throughout the world. As a critical component of sustainable
development, transportation systems should be designed to make positive contributions to the economic, environmental, and social
sustainability of the served regions and communities. This requirement significantly uplifts the challenges on the modeling and the
analysis of NDP. In this paper, we provide a review on the sustainable road NDP. Specifically, an overview on the three dimensions of
sustainable development (i.e., economic, environmental, and social) is first provided, focusing on their representative performance
measures relevant to road NDP. Then, we review the existing studies with the classification system of economy and environment-
oriented sustainable NDP, economy and equity-oriented sustainable NDP, and three-dimensional sustainable NDP. Future research
directions are suggested for advancing the methodological advancement and practical applications of sustainable transportation NDP.
Keywords: Sustainability, transportation, network design problem, multi-objective, bi-level programming
··································································································································································································································

1. Introduction Nitrogen Oxides (NOx), ozone (O3), sulfur dioxide (SO2),


Particulate Matter (PM), and other hazardous air pollutants). For
Sustainability development is a major concern all over the example, motorized vehicles in the United States contribute to
world, but it is not a new concept. It is typically defined as the 86% of CO emissions, 62% of NOx emissions, 12% of PM2.5
development that “meets the needs of the present without emissions, and 7% of SO2 emissions (Environmental Protection
compromising the ability of future generations to meet their own Agency, 2008). The long-term negative impact of environmental
needs” according to the Brundtland Report (also entitled “Our degradation is the inability to sustain human life. As a critical
Common Future”) released by the United Nations World component of sustainable development, transportation systems
Commission on Environment and Development in 1987 (Brundtland should be designed to make positive contributions to the
Commission, 1987). Sustainable development is typically charac- economic, environmental, and social sustainability of the
terized in terms of three dimensions: “economic, environmental, served regions and communities. The importance of achieving
and social” or “ecology, economy, and equity”. Generally, economic sustainability in transportation systems cannot be over emphasized.
sustainability guarantees the economic viability of actions for To better achieve sustainable development in transportation, it is
meeting present needs and reserving resources for future generations. critical to address how planners should design a transportation
Environmental sustainability is mainly concerned with the natural system that not only enhances mobility and economic growth but
environment and maintaining its diversity and productivity. Social also minimizes environmental degradation and climate change
sustainability mainly concerns the concept of “needs”, particularly for both present and future generations.
the essential needs of the poor. To transform the above fundamental definition of sustainable
Transportation systems provide social and economic connections development into goals and objectives, a sustainable transportation
in our daily life. However, the mobility is balanced against the system should be able to meet the multifaceted increasing travel
economic, environmental, and social costs imposed by the demands derived from social and economic activities, to enhance
system. Transportation systems are large contributors to Greenhouse the environment and quality of life of humans, and to benefit all
Gas (GHG) emissions. Increased quantities of motorized vehicles people with heterogeneous social-economic characteristics.
in the future are expected to contribute greater amounts of Achieving a sustainable transportation system relies on the
pollutants to the atmosphere (e.g., noise, carbon monoxide (CO), ability to integrate multiple effective strategies (e.g., integrated

*Associate Professor, Key Laboratory of Road and Traffic Engineering, Tongji University, Shanghai 201804, China (E-mail: xiangdongxu@tongji.edu.cn)
**Professor, Key Laboratory of Road and Traffic Engineering, Tongji University, Shanghai 201804, China; Dept. of Civil and Environmental Engineering,
Utah State University, Logan, UT 84322-4110, USA (Corresponding Author, E-mail: anthony.chen@usu.edu)
***Professor, Key Laboratory of Road and Traffic Engineering, Tongji University, Shanghai 201804, China (E-mail: tongjiyc@tongji.edu.cn)

− 1084 −
A Review of Sustainable Network Design for Road Networks

land use-transportation development planning, improvement of the methodological advancement and the practical applications
public transportation services and green (or non-motorized) of sustainable transportation NDP. The remainder of this paper is
transportation infrastructures, implementation of advanced organized as follows. Section 2 provides an overview of the
technologies in transportation systems management and demand sustainability concept, focusing on the performance measures of
management, environmental protection policies, etc.). the three dimensions of sustainable development relevant to road
Various measurement, modeling, evaluation, and analysis NDP. Section 3 reviews the bi-level programming modeling
frameworks have been developed regarding the planning, design, framework of NDP and the existing studies on the sustainable
management, operation, and control strategies of a sustainable road NDP. Finally, Section 4 provides future research suggestions
transportation system from either the demand side or the supply on this topic.
side or both. For example, Lopez and Monzon (2010) integrated
sustainability issues in strategic transportation planning by 2. Three Dimensions of Sustainable Development
developing a multi-criteria (i.e., the three dimensions of
sustainability) model for assessing transportation infrastructure Sustainable development has typically been characterized in
plans in a GIS-based computer framework. Kamga et al. (2012) terms of three dimensions as shown in Fig. 1: “economic,
discussed the basic concepts of sustainability in transportation, environmental and social” or “ecology, economy and equity”,
identified key parameters and existing deficiencies in transportation also coined as the “three-legged stool” of sustainability. A
systems, and suggested ways to achieve more sustainable sustainable transportation system should contribute to the
development. Yim et al. (2011) developed an optimization economic, environmental and social sustainability of the served
model for the joint land use and transportation network design regions and communities. The increasing social and economic
problem with consideration of demand uncertainty. Szeto et al. connections require mobility increase of transportation systems.
(2012) provided a comprehensive review on network equilibrium However, the benefits of increased mobility need to be balanced
modeling and optimization approaches to environmental against the economic, environmental, and social costs imposed
sustainability. Maheshwari et al. (2015) proposed a dynamic by the system.
model for planning sustainable transportation systems via a There are many performance measures under each of the three
system of nonlinear differential equations representing the dimensions of sustainable development (e.g., Litman, 2014). In
dynamics of three independent states (i.e., transportation, this review, only the typical performance measures relevant to
activity, and environmental systems), and discussed a policy road NDP are considered. We should point out that the three
scenario regarding investment in energy-efficient technologies dimensions have overlaps. For example, all the economic costs
and its effects on the states. (e.g., wasting travelers' time and slowing the delivery of goods
Road Network Design Problem (NDP) is a strategic decision- and services), environmental costs (e.g., generating vehicular
making problem in planning, designing and managing road emissions and noise, and worsening the air quality), and social
networks. It aims to make efficient use of the limited resources costs (e.g., negative impacts of vehicular pollutant on public
(e.g., land use and monetary budget) for optimizing the performance health, and equity issues) of traffic congestion can be converted
of transportation networks, while explicitly considering the to monetary (economic) costs. In the interest of clarity, we
travelers’ behavioral responses (e.g., residential location choice, discuss these three dimensions individually. The economic
mode choice, and route choice) to the design scheme. Due to its dimension of sustainability is only referred to as the congestion
theoretical value and practical significance, modeling, algorithm cost.
development, and applications of NDP have been extensively
studied by engineers, mathematicians, operations researchers, 2.1 Economic Dimension
and planners in the past few decades. For a comprehensive
review on classical NDP studies, interested readers are directed 2.1.1 Total Travel Time
to Bell and Iida (1997), Boyce (1984), Current and Marsh Traffic congestion generates economic costs by wasting
(1993), Friesz (1985), Magnanti and Wong (1984), and Yang and
Bell (1998). In the literature, the dominant objective of NDP
studies is to alleviate traffic congestion, e.g., minimize total
travel time, and maximize network reserve capacity (Yang and
Bell, 1998). They may fail to adequately consider the environmental
and social impacts of transportation systems as well as their
conflicting effects. The promotion of sustainable development
and sustainable transportation significantly uplifts the requirements
and challenges on the modeling and analysis of NDP.
In this paper, we provide a review on the sustainable road
network design problem, and propose some future research Fig. 1. Three Dimensions of Sustainable Development (Source:
suggestions that are believed to be able to significantly expand https://en.wikipedia.org/wiki/Sustainable_development)

Vol. 20, No. 4 / April 2016 − 1085 −


Xiangdong Xu, Anthony Chen, and Chao Yang

travelers' time and slowing the delivery of goods and services. In terms of network reserve capacity, Gao and Song (2002)
Total Travel Time (TTT) is a widely used measure to quantify the extended the concept to consider O-D pair-specific demand
total economic cost or efficiency of the network. Mathematically, it multipliers, while Chootinan et al. (2005) and Chen et al. (2006)
can be expressed as follows: further developed it as a reliability-based NDP for a road
network and a signal-controlled road network, respectively. Yang
Total travel time = ∑ ta ⋅ va (1)
a∈ A et al. (2000) formulated the network capacity and level of service
problem by determining the maximum zonal trip generation
where A is the set of directed links on the network; ta and va are
subject to combined distribution and assignment equilibrium
(average) travel time and flow on link a. TTT is also the objective
constraints. Chen and Kasikitwiwat (2011) and Chen et al.
function of the System Optimal (SO) traffic assignment model.
(2013) further detailed the network capacity model of Yang et al.
Chen et al. (2007) extended the TTT measure to consider
(2000) as the ultimate and practical network capacity models in
uncertainty and developed the alpha reliable network design
assessing network capacity flexibility and network capacity
problem that minimizes the TTT budget required to satisfy the
reliability, respectively.
TTT reliability constraint.
Remark 1: Yang and Wang (2002) examined the level of
equivalence and effectiveness of optimizing total travel time and
2.2.2 Economic Benefit
network reserve capacity in continuous NDP. They found that
TTT is applicable for the fixed demand case. It is inappropriate
the level of equivalence of the two objectives varies with the
when considering demand elasticity (e.g., dependency of O-D
congestion level. When the congestion level is low, maximizing
demand on O-D cost or level of service). In the elastic demand
network reserve capacity has the simultaneous effect of total
case, we can use the net economic benefit or social welfare:
w
travel time minimization. When the congestion level is high,
Economic benefit = ∑ ∫ q D–w1( ω ) dω – ∑ ta ⋅ va (2) maximizing network reserve capacity can only minimize the
0
w∈W a∈A
total travel time to a certain extent.
where W is the set of O-D pairs; qw is the travel demand of O-D Remark 2: Santos et al. (2010) developed the following
pair w; D–w1 is the inverse demand function. The inverse demand weighted link reserve capacity measure to quantify the network-
function can be regarded as an amount that a traveler is willing to wide reserve capacity:
pay for his/her travel, or a benefit that he/she can obtain from this a

travel (Yang and Huang, 2005). With this interpretation, the ∑ ( C a – v a ) va L a ⁄ ∑ v a L a (6)
a∈ A a∈ A
economic benefit is the total user benefit minus the total social
where La is the length of link a; α is a weighting parameter. Note
cost.
that the denominator is the total vehicle miles traveled. Thus, this
measure is an aggregation of link reserve capacity Ca-va
2.1.3 Network Reserve Capacity
weighted by the relative proportion of vehicle miles traveled on
Alternatively, network reserve capacity can also be used to
this link. This weighing scheme implies that we pay more
quantify the economic sustainability. The concept of network
attention to the spare capacity on heavy-flow and long links. This
reserve capacity was originally proposed by Wong and Yang
measure is simple and easy to calculate. However, it only serves
(1997) for a signal-controlled road network. It was defined as the
as a proxy or a localized approximation of the network-wide
largest multiplier µ applied to the existing O-D demand matrix q
reserve capacity. In contrast, the network reserve capacity
that can be allocated to the network without violating a pre-
measure in Eq. (3) is a global or network-wide measure based on
specified Level of Service (LOS). Mathematically, finding network
an optimization-based approach that can explicitly determine the
reserve capacity multiplier µ can be formulated as:
maximum throughput while considering travelers’ route choice
max µ (3) responses to a change of travel demand.
s.t. va( µq ) ≤ θaCa, ∀a ∈ A (4)
2.2 Environmental Dimension
va( µq ) ⇐ user equilibrium (5) Environmental protection is another important concern in
transportation system decisions, especially when considering the
where θa is a parameter denoting the pre-specified LOS required
sustainability issue of transportation systems. The environmental
on link a; Ca is the capacity of link a; va(µq) is the flow on link a,
dimension mainly quantifies the impacts of vehicles-generated
which is obtained by solving the lower-lower User Equilibrium
pollutants (e.g., vehicular emissions and noise). In the literature,
(UE) assignment problem under a given µ. The largest value of µ
the focus has been devoted to the emission effect since it is the
indicates whether the network has reserve capacity or not: if µ < 1,
major part of the vehicular pollution contributing to the
the network is overloaded by 100(1-µ) percent of q; otherwise, it
environment deterioration. Szeto et al. (2012) recently provided
has a reserve (or spare) capacity amounting to 100(µ-1) percent
a comprehensive review on the existing emission models based
of q. Essentially it is equivalent to the maximum throughput of
on the classification of aggregated, static, and dynamic approaches.
the network max ∑ µqw = maxµ ⋅ ∑ qw ⇔ maxµ .
w∈W w∈W As for the indicators of transportation noise, Pronello and

− 1086 − KSCE Journal of Civil Engineering


A Review of Sustainable Network Design for Road Networks

Camusso (2012) provided a review on both the general acoustic transportation policies and actions (e.g., congestion pricing and
ones and those used for specific transport modes. traffic rationing) in order to justify their positive contributions to air
It is well known that CO, hydrocarbons (HC), and NOx are the quality improvement and its subsequent public health. Wang et
main types of vehicular emission pollutants. Among different al. (2014a) presented a two-step procedure to assess the negative
types of vehicular emissions, CO is usually regarded as an impact of vehicle emissions on health: estimate the pollutant
important indicator for the level of atmospheric pollution uptake by individuals on each used path, and assess the health
generated by vehicular traffic (e.g., Alexopoulos et al., 1993; Yin impact based on population exposure level. For the first step
and Lawphongpanich, 2006; Nagurney et al., 2010; Chen et al., (modeling the pollutant uptake by individuals during their trip),
2011b; Chen and Yang, 2012; Wang et al., 2014a; Xu et al., one can adopt a three-stage approach: (a) modeling link emission
2015). In addition, CO is the most critical pollutant among the rates based on traffic flow, average vehicle speed and vehicle
various types of vehicular emissions, and almost all CO fleet composition, (b) modeling air pollutant concentrations from
emissions in the air are emitted by vehicles. Furthermore, the the emission rates and surface meteorology, and (c) modeling
emission functions of other pollutants are similar to that of CO pollutant uptake by travelers from the air pollution concentrations,
(Li et al., 2012). The most widely used vehicular CO emission travel time along each link, and breathing rate of travelers. Based
function is as follows: on the above three-stage approach, one can obtain the population
exposure level (e.g., total CO dose) on each used path. Then, we
Total CO emission = ∑ ea( va ) ⋅ va (7) can apply the median or maximum operators to assess the health
a∈A
impact of pollutant uptake on road users over different used
where ea(va) denotes the amount of CO pollution from link a. paths in the network.
The following nonlinear macroscopic model of Alexopoulos et
al. (1993) is a representative functional form to estimate the link- 2.3.2 Equity Issues
based vehicular CO emission. Various definitions of equity have been suggested in the
literature. It generally refers to the fairness and justice of the
0.7962 ⋅ L
ea ( va ) = 0.2038 ⋅ ta ( va ) ⋅ exp ⎛ -------------------------a⎞ (8) distribution of impacts (benefits and costs) of an action on two or
⎝ t a ( va ) ⎠
more units (Santos et al., 2008). In transportation planning,
where La, ta and ea are measured in kilometers, minutes and equity mainly concerns the fairness of the right to access
grams per hour, respectively. For other types of pollutants (e.g., transportation infrastructures for different groups of people.
CO2 and NOx), researchers (e.g., Hizir, 2006) have developed Equity should be carefully assessed when planning, designing,
their total emission functions, which are similar to the above and managing transportation networks. Interested readers may
form but with different parameters. also refer to Levinson (2010) for a comprehensive review on the
Note that Eq. (8) is a non-decreasing function of link flow as equity effects of road pricing. According to the study subject of
long as the speed is lower than 75 km/h (Xu et al., 2015). equity issues, it can be classified into horizontal and vertical
However, in reality, the emission functions may not be necessarily equities, terminologies used in Szeto et al. (2012). Horizontal
non-decreasing. For example, Sugawara and Niemeier (2002) equity involves the even distribution of the negative impacts of
proposed the following CO emission function, which is neither traffic externalities (e.g., congestion and emissions) within the
increasing nor decreasing, neither convex nor concave. This same population group/class having the same abilities and needs
function was also used by Wang et al. (2014a) in the three- (e.g., social class, value of time, and income). Vertical equity
dimensional sustainable road pricing model. involves the even distribution of the negative impacts of traffic
2 externalities among various groups/classes of people. Equity has
ea ( sa ( va ) ) = exp [ –0.05627 ( sa – 16 ) + 0.002187 ( sa – 16 ) ,
(9) different specifications, e.g., spatial equity, environmental equity,
–5 3 –6 4
–7.609 ⋅ 10 ( sa – 16 ) + 1.225 ⋅ 10 ( sa – 16 ) + 1.837]La and intergeneration equity.
where sa is the average speed on link a, which is a function of Spatial equity: In the NDP context, Meng and Yang (2002)
link flow va according to link travel time function. sa and La are proposed a framework to study the spatial distribution impact of
measured in mile/h and mile, respectively. the change in a general NDP and measured the spatial equity
impact by the maximum ratio of O-D travel times after and
2.3 Social Dimension before capacity enhancement:
This section divides the social dimension into health impact
and equity issues. Equity = ∑ { π w ( x ) ⁄ π w ( 0 ) } (10)
w∈W

2.3.1 Health Impact where πw(x) and πw(0) are the minimum travel times of O-D pair
Air quality deterioration caused by Greenhouse Gases (GHG) w after and before implementing network design scheme x,
and pollutants of vehicular emissions have a negative impact on respectively.
public health. Assessing this health impact is particularly important In a more general sense, there are various equity measures in
to a successful acceptance and implementation of some the literature, as summarized in Table 1. Note that these measures

Vol. 20, No. 4 / April 2016 − 1087 −


Xiangdong Xu, Anthony Chen, and Chao Yang

Table 1. A Summary of Equity Measures (modified based on Ramjerdi is disproportionately and adversely impacted. They proposed
(2006) and Feng and Zhang (2014)) four objective functions according to the various interpretations
Equity measure Formulation of EJ.
Range Ymax−Ymin w w w w w w w w 2
⎛ ∑ sx q p ⁄ ex ∑ sx q ( 1 – p ) ⁄ ex ⎞
1
EJ1 = ⎜ ----------------------------- -⎟
n 2
Variance --- ∑ i = 1( Yi – Y) w
– -----------------------------------------
w
(11)
n ⎜ w w
q (1 – p ) ⎟
w w
⎝ w ∑ q p ∑ ⎠
w
1 n 2
Coefficient of Variation (CV) C = --- ∑ i = 1 ( Yi – Y) ⁄ Y
n w w w w w w w w 2
⎛ ∑ ( sx q p ⁄ e x – s 0 q p ⁄ e0 )
1 n
--- ∑ i = 1 Yi ⁄ Y – 1 EJ2 = ⎜ ----------------------------------------------------------------
w
-
Relative mean deviation ⎜ w w
n ⎝ ∑q p
w
(12)
1 n
--- ∑ i = 1 logY – logYi w w w w w w w w 2
Mean logarithmic deviation ∑ s q ( 1 – p ) ⁄ ex – s0 q ( 1 – p ) ⁄ e 0 ⎞
x
- ⎟⎟
n
– ----------------------------------------------------------------------------------
w
w w
1 n 2 ∑ q ( 1 – p ) ⎠
Logarithmic variance --- ∑ i = 1( log ( Yi ⁄ Y) ) w
n
w w w w w w 2
1 n
--- ∑ i = 1( log ( Yi ⁄ Ylog ) )
2 ⎛ ∑ s x q p ∑ s x q ( 1 – p )⎞
Variance of logarithms
n EJ3 = ⎜ ---------------------
w w
-⎟
– --------------------------------- (13)
⎜ qw pw q (1 – p ) ⎟
w w

1- n n ⎝ w ∑ ∑ ⎠
Gini ---------- Y – Yj
2 ∑ i = 1∑ j = 1 i
w

2n Y
w w w w w w 2 w w w w w w 2 2
1 n
⎛ ∑ ( sx q p – s 0 q p ) ∑ ( sx q ( 1 – p ) – s0 q ( 1 – p ) ) ⎞
Theil’s entropy --- ∑ i = 1 ( Yi ⁄ Y)log ( Yi ⁄ Y)
n EJ4 = ⎜⎜ ------------------------------------------------
w
w w w w
-⎟
- – -------------------------------------------------------------------------
w

⎝ ∑ q p ∑q (1 – p ) ⎠
w w
1⁄1–ε
1 n
1 – ⎛⎝ --- ∑ i = 1 ( Yi ⁄ Y) ⎞⎠ (14)
1–ε
( ε ≠ 1)
n w w
Atkinson where s and s are the shortest travel time of O-D pair w
0 x
⎛ 1 – exp ⎛ 1
---
n
log ( Yi ⁄ Y)⎞ ⎞ ( ε = 1 ) without and with NDP scheme x; ew0 and ewx are the shortest
⎝ ⎝n∑ i = 1 ⎠⎠
distance of O-D pair w without and with NDP scheme x; pw is the
1 1 n
Kolm ---log ⎛ --- ∑ i = 1 exp ( α ( Y – Yi ) )⎞ proportion of trips between O-D pair w made by the protected
α ⎝n ⎠ population. In general, these EJ measures quantify the distribution
Notation: Yi is a measure of welfare/accessibility; n is the number of of travel times amongst population groups.
observations on welfare/accessibility; Y is the mean of Yi; Ylog is the Intergeneration equity: Sustainable development is the
mean of logYi; ε and α in Atkinson and Kolm measures are positive
parameters of addressing inequity aversion. development that can meet the needs of both the present and
future generations. Szeto and Lo (2006) examined the issue of
intergeneration equity according to the user and social
are flexible in the sense that Y can have different specifications, e.g., perspectives by using a time-dependent NDP framework of
welfare (net economic benefit) or accessibility. Also, to cater for planning the optimal infrastructure improvement timetable, the
the case with multiple O-D pairs, O-D demands can be considered associated financial arrangement, and tolling scheme over the
in n and also as a weight of Y. planning horizon. The social intergeneration equity was
Environmental equity: Rilett and Benedek (1994) studied the measured by the total social surplus per capita discounted for the
traffic assignment under environmental and equity objectives. time effect, representing the gross average net benefit of the
The expected exhibition of this traffic assignment problem is that transportation infrastructure projects on society from generation
vehicles are routed through the network in such a way that no to generation. The user intergeneration equity was measured by
one group of people living near the network is affected more than link tolls and discounted O-D travel costs. A gap function was
any other groups of people. In other words, negative externalities defined to measure the degree of intergeneration equity achieved
of traffic (e.g., the amount of emissions or noise released on all over generations.
streets or adjacent streets) are distributed as evenly as possible
Intergeneration Equity = ∑ Ni =–11∑ Nj = i + 1 wi – wj 2 (15)
among competing routes, and do not exceed a certain threshold
for health. where N is the number of design periods; wi and wj denote the
Environmental justice (Schweitzer and Valenzuela, 2004) is a vector of equity measures (e.g., total discounted social surplus
closely relevant concept with environmental equity. It can be per capita, link tolls, and discounted O-D travel costs) at periods i
regarded as an extension of environmental equity but focusing and j. This gap function is always non-negative, and has a lower
on minority and low-income population. Duthie and Waller value for more equitable schemes. Alternatively, Szeto et al.
(2008) incorporated the Environmental Justice (EJ) consideration (2015) used the sum of the variance of discounted generalized
into discrete NDP. EJ means that no protected population group user costs of all O-D pairs as a measure of intergeneration equity.

− 1088 − KSCE Journal of Civil Engineering


A Review of Sustainable Network Design for Road Networks

3. Overview of Sustainable Road NDP link toll rate, etc. The examples of discrete decision vari-
ables include adding new roads/lanes, deleting existing links
3.1 Modeling Framework of Network Design Problem (i.e., road closure), configuration of one-way streets and
Road NDP involves the interaction of the two main decision- contraflow lanes, etc. The mixed NDP is a combination of
making stakeholders: network planners and travelers. Travelers continuous and discrete NDPs, e.g., cordon-based pricing,
selfishly make their travel choice decisions (e.g., mode/route adding new infrastructure and capacity expansion, simulta-
choice) according to their own criteria (e.g., shortest travel time, neous location-network design problem, etc. Nowadays
lowest travel cost, etc.) and knowledge of the network state; there are hardly any feasible sites for further expansion of
while planners aim to make efficient use of limited resources to existing road networks, especially in large Asian cities. In
achieve the stated objectives/requirements. When evaluating a this case, the NDP needs to focus on how to make efficient
network design scheme before implementation, planners need to use of the existing network (instead of adding new roads or
know how travelers respond to this scheme, i.e., the aggregation expanding link capacity) while enhancing the network sus-
of individual travelers’ travel decisions in terms of mode-specific tainability. Signal timing, ramp metering, configuration of
travel demand pattern and traffic flow pattern. Planners’ design one-way streets and contraflow lanes, and road pricing are
decisions can affect (but not directly determine) travelers’ travel some of the available design schemes. These design schemes
behaviors and the distribution of traffic flows, whose aggregation have recently been considered with different dimensions of
will in turn influence planners’ design decisions. Hence, travelers’ sustainability, e.g., congestion/emission pricing (Maruyama
response to a NDP scheme should be considered in the NDP. and Sumalee, 2007; Subprasom and Chen, 2007; Sumalee et
With this consideration, the NDP decision-making process is al., 2009; Ho and Sumalee, 2010; Friesz et al., 2013; Sharma
usually modeled as a bi-level programming (BLP) structure or a and Mishra, 2013; Wang et al., 2014a; Amirgholy et al., 2015;
non-corporative game between two players (planners as leaders Ma et al., 2015), dynamic traffic management measures (Wis-
and travelers as followers). mans et al., 2011), and signal timing (Li and Ge, 2014).
Mathematically, the BLP can be expressed as follows: * Objectives: According to the NDP specification, the objec-
tive function could be total travel time minimization, net-
lim F ( x, y( x ) )
[Upper level] min work reserve capacity maximization, total vehicular emission
x∈X
subject to G(x, y(x) ≤ 0 minimization, etc. The NDP usually involves a benefit game
is a solution of the following optimization for an fixed among different stakeholders. Each stakeholder has his/her
where y(x)
x∈X .
own requirement on the NDP decision, rendering it as a
lim f ( x, y )
[Lower level] min
y∈Y multi-objective decision problem. Some of these objectives
subject to g(x, y) ≤ 0
are even conflicting. In other words, improving one objec-
In the above formulation, x and y are decision vectors of the tive may worsen one or more of the other objectives.
upper-level and lower-level subprograms, respectively; X and Y * Constraints: According to the NDP specification, the con-
are the domains of x and y, respectively; F and G are the straints could be link capacity constraint, environmental
objective(s) and the constraint set of the upper-level subprogram, capacity constraint, land use constraint, construction budget
and f and g are the objective(s) and the constraint set of the constraint, network performance constraint, etc.
lower-level subprogram. The upper-level subprogram describes * Travel choice behaviors: In the lower-level subprogram,
the planners’ decision problem, and the lower-level subprogram typically we use traffic equilibrium assignment problem to
describes the travelers’ behavior problem. Note that the decision obtain the equilibrium traffic flow pattern under the plan-
vector y of the lower-level subprogram is a function of the design ners’ specified NDP scheme. According to the travel choice
vector x of the upper-level subprogram. Generally, there is no dimensions and choice criteria, the lower-level subprogram
explicit relationship between x and y. Therefore, BLP is usually could be deterministic UE (DUE), DUE with capacity con-
quite difficult to solve. Mathematically, it is generally nonlinear straints, DUE with elastic demand, Stochastic User Equilib-
and nonconvex. rium (SUE), multi-class network equilibrium models, non-
Below we briefly summarize the four components of the NDP additive traffic equilibrium models, combined travel demand
as a BLP. models, combined travel demand models with land use, net-
* Decision variables: According to the mathematical type of work equilibrium under uncertainty, etc.
decision variables, the NDP has three categories: continuous Recently, there are two latest comprehensive reviews on
NDP, discrete NDP, and mixed NDP. Generally, the continu- transportation NDP with different focuses.
ous NDP involves the parameterization of the network while * Chen et al. (2011a) focused on the uncertainty modeling in
the discrete NDP involves the topology of the network. Due transportation NDP and provided some new developments
to the continuity of decision variables, majority of NDP on a bi-objective reliable NDP model that explicitly opti-
studies in the literature belongs to the continuous NDP. The mizes the tradeoff between capacity reliability from the sup-
examples of continuous decision variables include link ply side and travel time reliability from the demand side.
capacity enhancement, signal timing, ramp metering rate, * Farahani et al. (2013) focused on the Urban Transportation

Vol. 20, No. 4 / April 2016 − 1089 −


Xiangdong Xu, Anthony Chen, and Chao Yang

Network Design Problem (UTNDP) related to urban trans- algorithms have been widely studied in the literature, which will
portation network topology and its configuration. Defini- not be reviewed in this section. Also, economic dimension is
tions, general formulations and classifications of UTNDP typically considered in the environment-oriented or equity-oriented
(including road NDP, transit NDP, transit NDP and fre- NDPs. Hence, the studies reviewed in this section consider at
quency setting problem, multi-modal NDP), solution meth- least two dimensions of sustainability. According to the dimensions
ods, and applications in real-world case studies were of sustainability considered in the NDP, we classify the existing
comprehensively reviewed. studies into three groups: economy & environment-oriented
sustainable NDP, economy & equity-oriented sustainable NDP,
3.2 Sustainable Road Network Design Problem and three-dimensional sustainable NDP. Below we summarize
Note that the single economy-oriented NDP models and selective studies in each group.

Table 2. A summary of Existing Studies on Economy & Environment-oriented Sustainable NDP (not reviewed in Szeto et al. (2012))
Sustainability Travelers’ behavioral
Reference Decision Sustainability measure Solution method
dimension model
· Total travel time
Sharma and Capacity · Economy · Total emissions (speed-dependent · Upper level: NSGA II
UE
Mathew (2011) expansion · Environment emission functions for various · Lower level: Frank-Wolfe
modes and pollutants)
· Total travel time · Multi-objective evolu-
Dynamic traffic · Economy
Wismans et al. · Total CO2 emissions tionary algorithms
management · Climate Dynamic UE
(2011) · Average weighted sound power (NSGAII, SPEA2 and
measures · Noise
level at source SPEA2+)
Chen and Yang · Economy · Total travel time
Toll and rebate UE /
(2012) · Environment · Total CO emissions
· Total travel time
Duell et al. Capacity · Economy · Total energy consumption (internal
UE NSGA II
(2013) enhancement · Environment combustion engine vehicles and
plug-in electric vehicles)
Dynamic UE (travel cost
Friesz et al. Dynamic · Economy · Total travel cost includes travel time, Quadratic penalty-based
(2013) congestion pricing · Environment · Total emissions early/late arrival penalty gradient projection
and toll)
· Total travel time
· Total energy consumption (internal
Gardner et al. Capacity · Economy combustion engine vehicles and
UE /
(2013) enhancement · Environment plug-in electric vehicles)
· Total emissions (CO2, VOC, SO2
and NOx)
Sharma and Mishra Emission · Economy · Total travel time UE/combined modal · Upper level: GA
(2013) pricing · Environment · Total system GHG emissions split-traffic assignment · Lower level: Frank-Wolfe
· Cost · Total construction cost · Upper level: GA based on
UE under each
Xu et al. Capacity · Safety · Total number of incidents random simulation and
realization of uncertain
(2013) enhancement · Efficiency · Total travel time fuzzy evaluation
travel demand
· Environment · Total CO emissions · Lower level: Frank-Wolfe
Capacity expansion,
· Consumer surplus and producer Penalty function method +
employment Scenario-based
· Economy surplus simulated annealing +
Li et al. (2014) development & stochastic location and
· Environment · Damage cost of total traffic emis- Gauss-Seidel decomposi-
residential develop- route choice equilibrium
sions tion
ment
· Upper level: Enhanced
· Total travel time
chemical reaction opti-
Szeto et al. Discrete capacity · Economy · Total emission cost (CO, VOC
UE mization (weighted objec-
(2014) addition · Environment and NOx)
tive)
· Total noise cost
· Lower level: Frank-Wolfe
· Upper level: Multi-
· Consumers' surplus Joint trip production-
attribute decision method
Amirgholy et al. CBD cordon · Efficiency · Total travel time distribution-mode
(Electre)
(2015) price · Environment · Total CO emissions choice-assignment
· Lower level: Diagonal-
· Migration level of retailers model
ization
· Total travel and waiting time Decomposed and solved
Dynamic emission · Economy
Ma et al. (2015) costs Dynamic UE as two separate subprob-
toll · Environment
· Total CO emissions lems
A modified NSGA-II; a
Miandoabchi et al. Both tactical and · Economy · Total travel time
UE multiobjective B-cell algo-
(2015) strategic decisions · Environment · Total CO emissions
rithm

− 1090 − KSCE Journal of Civil Engineering


A Review of Sustainable Network Design for Road Networks

3.2.1 Economy & Environment-Oriented Sustainable NDP Optimization (PSO), Ant Colony Optimization (ACO), Chemical
Table 2 summarizes existing studies on the economy and Reaction Optimization (CRO), non-dominated sorting genetic
environment-oriented sustainable road NDP. Herein we highlight algorithm (NSGA-II), Strength Pareto Evolutionary Algorithm
the application/decision of the NDP, sustainability dimensions (SPEA2), B-cell algorithm, interchange algorithm, and variable
considered and the corresponding performance measures, travelers’ neighborhood search algorithm. Besides the meta-heuristics,
behavioral model (i.e., lower-level model), and solution method. some studies also use analytical approaches for the solution, e.g.,
We should point out that Szeto et al. (2012) also provided a quadratic penalty-based gradient projection, sensitivity-analysis
summary table on the bi-level transportation problem with based method, generalized reduced gradient method, partial
environmental considerations. To avoid repetition, Table 2 only linearization method, finite element & Newtonian-based algorithm,
reviews the more recent studies that are not included in Szeto et derivative-free algorithms, relaxation and decomposition methods,
al. (2012) and adds more information on their lower-level and multi-attribute decision method (Electre). Note that, other
models and economy measures. than emission pricing, a marketable pollution permit system
From Table 2, one can find that the decision problems of these (Nagurney et al., 1998; Nagurney, 2000) has also been proposed
studies are quite diverse, including the traditional continuous to reduce vehicular pollution and to achieve environmental
capacity enhancement and discrete capacity addition, dynamic quality standard.
traffic management measures (signal control and variable
message sign for changing speed limit), toll and rebate, static and 3.2.2 Economy & Equity-Oriented Sustainable NDP
dynamic congestion pricing, static and dynamic emission pricing, Table 3 summarizes existing studies on road NDPs with both
joint decision of capacity expansion, employment development and economy and equity considerations. The decision/application
residential development, and both tactical and strategic decisions problems of these studies are concentrated on the continuous
(sequence of link construction, expansion projects over a capacity enhancement and toll rate (majority), the discrete
predetermined planning horizon, configuration of street orientations, capacity addition, the road level and cordon selection, and their
and lane allocations for morning and evening peaks in each year combinations of discrete and continuous counterparts. Different
of planning horizon). As for the performance measures, TTT is from the domination of TTT in Table 2, social welfare is the
the most widely used one for the economy dimension. In most widely used performance measure of economy dimension
addition, more factors such as waiting time and early/late arrival in Table 3. The other performance measures are quite diverse,
penalties are also considered based on TTT; consumer surplus including total travel time, total trip generations, profit of
and producer surplus (i.e. total revenue from housing rents) have private investors, aggregate accessibility to population
also been considered as measures of economy dimension. For centers, network reserve capacity, etc. As to the performance
the environmental dimension, the majority of these studies use measure of equity dimension, most studies use the spatial
the total vehicular emission as the performance measure. Among equity measure (i.e., the maximum change of generalized O-
different types of emission pollutants, CO and CO2 receive the D costs or link travel times or consumer surplus, for various
most attention, while NOx, VOC and SO2 receive less consideration. user classes with different values of time), or the Gini and Theil
It is worth noting that noise and energy consumption (considering indicators of user benefit/accessibility/user utility/expected
both internal combustion engine vehicles and plug-in electric equivalent income. The other equity measures are accessibility
vehicles) are receiving increasing attention. Other than the direct of a percentage (e.g., 20%) of centers with lower accessibility,
calculation of total vehicular emissions/noise, their cost form has environmental justice (in terms of four measures on distribution of
been used in order to have a consistent scale or unit with other travel times amongst population groups, or the Theil entropy of
considered dimensions. Most of these studies treat the problems environmental impact), intergeneration user/social equity, etc.
as deterministic, except for Xu et al. (2013) with planners’ When considering uncertainty issues, these objectives are modeled
subjective uncertainty regarding the linguistic setting of goals in the form of either expectation or chance constraint.
and priority structure and the objective travel demand uncertainty Similarly, the lower-level models are quite diverse, including
and Li et al. (2014) with uncertainty in future population. In UE and its various extensions (e.g., elastic demand, multiple
terms of the lower-level traveler’s behavioral model, the UE classes with different Values of Time (VOTs), time-dependent,
principle is the modeling foundation along with more considerations continuum network, and multi-modal), Probit SUE, trip chain-
such as dynamics, uncertainty, multiple choice attributes (e.g., based equilibrium, and combined models. As to the solution
toll, early/late arrival penalties, etc.), and combined models. To algorithm, both analytical and meta-heuristic algorithms
solve the NDP models with both economy and environmental mentioned above have been adopted.
considerations, most of these studies adopt or enhance different
meta-heuristics to directly solve the multi-objective problems or 3.2.3 Three-Dimensional Sustainable NDP
to solve the weighted-sum form/constrained-single-objective Compared to the above two categories, there are very limited
form/goal-programming form. Some of these meta-heuristics NDP studies that simultaneously consider all three dimensions of
include Genetic Algorithm (GA), Simulated Annealing (SA), sustainability as shown in Table 4. One can see that these studies
Artificial Bee Colony (ABC) algorithm, Particle Swarm have different design variables: capacity enhancement, toll,

Vol. 20, No. 4 / April 2016 − 1091 −


Xiangdong Xu, Anthony Chen, and Chao Yang

Table 3. A summary of existing studies on economy & equity-oriented sustainable NDP


Sustainability Travelers’ behavioral
Reference Decision Sustainability measure Solution method
dimension model
Meng and Yang Capacity · Economy · Total travel time Penalty function + sim-
UE
(2002) enhancement · Equity · Spatial equity ulated annealing
· Social welfare
· Economy · Maximum relative increase of general- Multiclass UE with
Yang and Zhang Penalty function + sim-
Toll · Spatial and ized O-D costs between all O-D pairs elastic demand and
(2002) ulated annealing
social equity for various user classes with different different VOTs
VOTs
· Upper level: simulation
· The expectation of total travel time UE under each realiza-
Chen and Yang Capacity · Economy based GA
· Spatial equity (expectation or chance tion of uncertain travel
(2004) enhancement · Equity · Lower level: gradient
constraint form) demand
projection
Connors et al. · Economy · Social welfare Sensitivity-analysis
Toll Probit SUE
(2005) · Equity · Theil equity constraint based method
Upper level: GA
Lee et al. Land-use · Economy · Total trip generations Combined trip distribu-
Lower level: convex
(2006) development · Equity · Spatial equity constraint tion/assignment UE
combination
· Total discounted social surplus
Capacity
Szeto and Lo · Economy · Gap function (in terms of intergenera- Time-dependent UE Generalized reduced
expansion and
(2006) · Equity tion user equity and social equity) con- with elastic demand gradient method
toll
straint
· Profit of private investors UE with elastic demand
Chen and Subprasom · Economy Simulation based
BOT link tolls · Social welfare under each realization of
(2007) · Equity multi-objective GA
· Gini coefficient uncertain travel demand
Maruyama and Cordon and area · Economy · Social welfare Trip chain-based equi- Partial linearization
Sumalee (2007) pricing · Equity · Gini coefficient of user benefit librium method
· Upper level: Selector-
· Economy · Four travel time-based measure
Duthie and Waller Discrete ecombinative GA
· Environmental · Four measures on distribution of travel UE
(2008) link addition · Lower level: Frank-
justice times amongst population groups
Wolfe
· Aggregate accessibility to population
centers Not applicable (long-
Santos et al. Road level · Economy · Three accessibility-based equity mea- term interurban road
/
(2008) decision · Equity sures: accessibility of a percentage of NDP, not modeled as
centers with lower accessibility, Gini BLP)
and Theil
· Weighted average accessibility Add plus interchange
Not applicable (long-
· Efficiency · Weighted reserve capacity algorithm; variable
Santos et al. Road level term interurban road
· Robustness · Accessibility to 20% of traffic genera- neighborhood search
(2009) decision NDP, not modeled as
· Equity tion centers with the lowest accessibili- algorithm; enhanced
BLP)
ties GA
· Social welfare improvement
Sumalee et al. Cordon congestion · Economy · Revenue generation Dynamic self-adaptive
SATURN model
(2009) pricing · Equity · Gini coefficient of net social welfare penalty GA; NSGA-II
improvement
Lo and Szeto Capacity · Economy · Total discounted consumer surplus Annual-based UE with Generalized reduced
(2009) expansion and toll · Equity · Spatial equity in consumer surplus elastic demand gradient method
· Economy · Social benefit
Cordon · Utility equity · Theil entropy for user utility Finite element method+
Ho and Sumalee
congestion · Environmen- · Theil entropy for environmental impact UE (continuum network) Newtonian-based algo-
(2010)
pricing tal justice · Logit based on utility that governs the rithm
· Acceptability acceptance
Wu et al. · Economy · Social benefit Derivative-free
Toll and credit Multimodal UE
(2012) · Equity · Gini index of expected equivalent income algorithms
· Equity Six accessibility-based equity indicators · Upper level: NSGA II
Feng and Zhang Capacity
· Construction (GINI, Theil, mean log deviation, rela- UE · Lower level: Frank-
(2014) enhancement
cost tive mean deviation, CV, and Atkinson) Wolfe
· Total travel time
· Spatial equity constraint in terms of UE with generalized
Wang et al. Capacity · Economy
link travel times travel cost (travel time Relaxation algorithm
(2014b) enhancement · Equity
· Spatial equity constraint in terms of O-D + credit charging)
travel costs

− 1092 − KSCE Journal of Civil Engineering


A Review of Sustainable Network Design for Road Networks

Table 4. A Summary Of Existing Studies on Three-dimensional Sustainable NDP


Sustainability
Reference Decision Sustainability measure Travelers' behavioral model Solution method
dimension
· Economy · Total travel time ·Upper level: simulation based
Chen and Xu Capacity UE under each realization
· Environment · Total CO emissions genetic algorithm
(2012) enhancement of uncertain travel demand
· Equity · Spatial equity · Lower level: Frank-Wolfe
· Change in consumer surplus
Time-dependent land-use
· Total CO emissions · Upper level: artificial bee
transportation problem (time-
Toll and · Economy · Landowner inequity (variance of colony algorithm
Szeto et al. dependent Lowry-type con-
capacity · Environment discounted landowner profit) and · Lower level: method of
(2015) straints and time-dependent
enhancement · Society/equity intergeneration inequity (variance successive averages and
modal split/assignment con-
of discounted generalized user cost Frank-Wolfe
straints)
over time)
· Total number of car
ownership and trips
· Upper level: multi-objec-
· Distributional differences between
Feng and · Mobility tive genetic algorithm
Zonal car zonal accessibilities (Gini) Combined distribution
Timmermans · Equity (NSGA-II)
ownership · Environmental capacity constraints and assignment model
(2014) · Environment · Lower level: Frank-Wolfe +
(link emission is calculated as product
Hitchcock problem
of link length, flow and average speed -
dependent emission factors)
· Network reserve capacity
· Economy
Li and Ge · Total CO emissions (move +stop UE with vehicle delays at Penalty function + simulated
Signal timing · Environment
(2014) delay) signalized intersections annealing (weighted sum)
· Equity
· Spatial equity constraint
· Economy · Total travel time Time surplus maximization
Wang et al. · Upper level: NSGA-II
Road pricing · Environment · Total CO emissions bi-objective (time & toll)
(2014a) ·Lower level: quasi-Newton
· Health impact · Level of pollutant uptake UE
· Revenue - construction cost, total
Toll and · Economy · Upper level: simulated
Yin et al. travel cost
capacity · Environment UE annealing
(2014) · Total CO emissions
enhancement · Equity · Lower level: Frank-Wolfe
· Spatial equity
· Network reserve capacity (mean-
variance)
· Upper level: particle swarm
· Economy · Link emission chance constraint
Wang et al. Capacity Bi-modal reliability-based optimization
· Environment · Demand equity chance constraint (at
(2015) enhancement SUE · Lower level: method of
· Equity least a certain percentage of increase
successive averages
of allowable travel demands for all O-
D pairs)

zonal car ownership, or signal timing. Total travel time and philosophies: expected value goal programming (GP) model,
network reserve capacity are the main measures for the economy chance-constrained GP model, and dependent-chance GP model
dimension; total CO emission (but with different emission in Chen and Xu (2012); mean-variance model of network
functions as in Wang et al. (2014a)) is the main measure for the reserve capacity in Wang et al. (2015). Given the multi-objective
environment dimension. As for the social dimension, the BLP structure, some studies treat the multiple objectives directly
performance measures include the maximum change of O-D by adopting a goal programming approach or converting the
travel time representing spatial equity in Chen and Xu (2012), Li multiple objectives to a weighed-sum single objective problem,
and Ge (2014) and Yin et al. (2014), the variance of discounted while other studies treat some objectives as constraints, and then
landowner profit representing landowner inequity and the use different types of metaheuristic algorithms for determining
variance of discounted generalized user cost over time representing the solution.
intergeneration inequity in Szeto et al. (2015), the Gini coefficient
of zonal accessibilities representing spatial equity in Feng and 4. Future Outlook
Timmermans (2014), the level of pollutant uptake representing
the negative health impacts in Wang et al. (2014a), and the After reviewing the literature of sustainable road NDP, we
demand equity chance constraint (at least a certain percentage of realize that there are some research gaps to be further addressed.
increase of allowable travel demands for all O-D pairs) in Wang In this section, we propose future research suggestions and
et al. (2015). To cater for different decision problems, the lower- directions that are believed to be able to significantly advance the
level subprogram has different specifications such as UE, bi- state-of-the-art and the state-of-practice of sustainable transportation
objective UE, or combined models with different considerations. NDP. These future research suggestions and directions include
Chen and Xu (2012) and Wang et al. (2015) explicitly model three main aspects: modeling realism, algorithm development,
demand uncertainty in the NDP with different modeling and practical applications.

Vol. 20, No. 4 / April 2016 − 1093 −


Xiangdong Xu, Anthony Chen, and Chao Yang

4.1 Modeling Realism sustainable NDP. The research questions in the time-depen-
* Uncertainty issues: Uncertainty exists in almost all aspects dent sustainable NDP include the developmental schedule
of our daily life, and transportation systems are no excep- and financial arrangement over time (e.g., 20 to 30 years
tion. Typical sources of uncertainty in transportation systems with each generation ranges from 5 to 10 years), how much
include travel demand fluctuation, link capacity degradation, to develop (e.g., all or incremental), who should pay for the
travelers’ imperfect perception of network conditions, and development, the intergeneration equity issue of the present
limited rationality of travelers’ choice behaviors. For exam- and future generations, etc.
ple, Ng and Lo (2013) modeled vehicular emissions by * Competition and cooperation of different spatial levels: To
introducing the conformity probability to capture the full achieve a win-win situation in the sustainable transportation
probabilistic behavior of vehicular emissions under capacity NDP, it is necessary to explicitly analyze and regulate the
and link flow uncertainties. The conformity probability was competition and cooperation among different management
defined as the probability of the maximum emission across authorities, different spatial regions/cities, and different
all links below a critical level. The stochastic dependencies transportation infrastructures within a city/region (e.g., ele-
were modeled using copulas, which generalized other com- vated roads and surface streets; expressway, arterial and
mon dependence modeling techniques in transportation net- local streets).
work modeling. In addition, there are some unique uncertainties * Travelers’ behavior model with sustainability consideration:
associated with the sustainable transportation NDP. For With public awareness of sustainability, travelers gradually
example, the impacts of budget uncertainty (e.g., due to eco- pay more attention to the externality of traffic congestion.
nomic fluctuation) and long-term travel demand uncertainty Hence, it is necessary to capture their multifaceted concerns
and differentiation (e.g., due to population aging and popula- (e.g., travel time, energy consumption cost, air quality,
tion policy adjustment like in China) on the achievement safety, health, etc.) in the lower-level travel choice and net-
level of sustainability need to be carefully addressed. Other work equilibrium models of the sustainable NDP model.
than the objective uncertainties mentioned above, we also This is particularly important in the bicycle network design
need to account for planners’ subjective uncertainties. A problem since route choice criteria for bicycles are signifi-
typical subjective uncertainty in the NDP decision is plan- cantly different compared to those for motorized vehicles.
ners’ target specification on the sustainability performance * Integration of three dimensions: Given the multi-objective
measures. The fuzzy goal programming approach of model- nature of the sustainable transportation NDP, it will be rea-
ing both subjective (fuzzy variables to model linguistic set- sonable to integrate and synthesize the three dimensions for
ting of goals and priority structure) and objective (random facilitating analysis and optimization. This integration is
variables to model travel demand uncertainty) uncertainties particularly useful in developing a practically operable sus-
simultaneously in the NDP developed by Xu et al. (2013) tainability index, and ranking the level of transportation net-
could be extended to tackle this issue. work sustainability among a set of cities or regions.
* Budget estimation problem: The budget estimation problem
(i.e., optimize the construction budget to satisfy a certain 4.2 Algorithm Development
performance requirement) can be considered as an inverse In general, the sustainable transportation NDP could be
problem of the traditional NDP (i.e., optimize the network formulated as a multi-objective BLP, where the multi-objective
performance under a given budget). Relative to the tradi- captures the three dimensions of sustainability and the BLP
tional NDP, the network-wide budget estimation problem structure considers planners’ decision and travelers’ behavioral
received very limited attention in the literature (Yin and response. Advanced techniques of solving multi-objective
Lawphongpanich, 2007; Yin et al., 2008), yet it has practical programming and BLP problems could be adapted to solve the
significance. The budget estimation in the sustainable NDP sustainable transportation NDP model in large-scale networks.
will be even more complicated. The challenge lies in the For example,
complex performance functions of the environmental and * Linearization techniques (e.g., Wang and Lo, 2010; Luathep
the social dimensions of sustainable transportation NDP, et al., 2011; Liu and Wang, 2015) could be adopted to refor-
which are being treated as constraints in the optimization of mulate the originally non-convex nonlinear BLP as a com-
network-wide budget estimation. putationally tractable mixed integer linear programming
* Time dimension: Recall that sustainable development is the (MILP) problem, whose solution possesses the desirable
development that can satisfy the needs of both the present property of global optimality.
and future generations. Transportation infrastructure con- * Alternatively, parallel computing techniques could be embed-
struction is a long-term and high capital investment. Also, ded into the meta-heuristics (e.g., genetic algorithm, simu-
for profit-driven investments, at the inchoate operating lated annealing, particle swarm optimization, etc.) for
stage, the infrastructure construction and operational costs enhancing the computational efficiency in large-scale net-
are not necessarily expected to be balanced annually. Hence, works. When applying the meta-heuristics, one of the com-
it is critical to explicitly model the time dimension in the plexities is the handling of constraints (e.g., specialized

− 1094 − KSCE Journal of Civil Engineering


A Review of Sustainable Network Design for Road Networks

sustainability requirement and budget constraint). A gradi- lower level included a car demand assignment model and a
ent-based repair method proposed by Chootinan and Chen bike demand assignment model. Pternea et al. (2015) devel-
(2006) could be adapted to systematically repair infeasible oped a model for the sustainable urban transit network
solutions by using the gradient information derived from the design problem by minimizing the operator, user, and envi-
constraint set. ronmental costs. This model explicitly incorporated environ-
* The sustainable transportation NDP is essentially a multi- mental impacts, considered emission-free (electric) vehicles,
objective optimization problem. Solving the multi-objective and introduced a direct route design approach with route
optimization models directly requires generating a family of structure and directness control. The design variables included
optimal solutions known as the Pareto-optimal set (Chen et route structures, frequencies and vehicle types (conven-
al., 2006, 2010). However, planners only need a good/satis- tional/ electric).
factory solution that meets the goals/targets of different * Public education of sustainable transportation NDP should
stakeholders in practical implementation. With this consid- be emphasized in order to receive more participation and
eration, it is meaningful to adopt the goal programming support in the decision-making process, and to enhance its
(GP) approach (Chen and Xu, 2012; Xu et al., 2013; Yin et acceptability by the public. Emerging technologies could be
al., 2014) for the solution. The GP approach is able to used herein. For example, a web-based GIS platform, simi-
explicitly consider the planners-defined goals and priority lar to Transitmix for transit route design (www.transit-
structure among the multiple dimensions in the sustainable mix.net), could be developed for the sustainable transportation
transportation NDP decision-making process. It determines NDP. The public can visually design/edit the network
a single good solution that can be readily used for imple- according to their own thinking and preference (e.g., a new
mentation, while best satisfying the set of planners-specified road, a new lane, a bus stop, a one-way street, a contraflow
goals. lane, parking lots, and an electric vehicle charging station
near home) in the platform. Then, the platform will assess
4.3 Practical Applications the three-dimensional sustainability performance from the
* Well-prepared test networks, scenarios sets, and realistic perspectives of both users themselves and the network as a
case studies using large-scale networks should be developed whole. This information will also enhance the understanding
for examining the practical applicability of the modeling of the public on the system externality with respect to net-
methodologies and solution techniques of sustainable trans- work sustainability.
portation NDP. For example, Steininger et al. (2007) ana-
lyzed the impacts of nationwide car road pricing schemes Acknowledgements
with respect to economic, environmental, and social indica-
tors of sustainability and quantified the trade-offs among the This study was supported by National Natural Science
three dimensions under different charging principles and Foundation of China (51408433), the Chang Jiang Chair
revenue recycling options. They employed a computable Professorship Program sponsored by the Ministry of Education
general equilibrium approach for the analysis and provided in China, and the Program for Young Excellent Talents in Tongji
detailed empirical analysis for the case of Austria. University.
* The sustainable transportation NDP should be expanded to
more decision-making domains. For example, it is straight- References
forward to consider a multi-modal network with park-and-
ride for providing more solutions to the private car domi- Alexopoulos, A., Assimacopoulos, D., and Mitsoulis, E. (1993). “Model
nated society. Also, the electric vehicle charging infrastruc- for traffic emissions estimation.” Atmospheric Environment Part B,
ture (or battery swapping station) location problem could be Vol. 27, No. 4, pp. 435-466, DOI: 10.1016/0957-1272(93)90020-7.
Amirgholy, M., Rezaeestakhruie, H., and Poorzahedy, H. (2015). “Multi-
evaluated and optimized from the perspectives of efficiency,
objective cordon price design to control long run adverse traffic
environmental, social, and financial sustainability. A road effects in large urban areas.” Netnomics, Vol. 16, No. 1, pp. 1-52,
diet is an effective way to maximize the utility for cyclists DOI: 10.1007/s11066-015-9092-9.
and to minimize the negative effect on motorists. Recently, Bell, M. G. H. and Iida, Y. (1997). Transportation network analysis, UK:
Sohn (2011) proposed a multi-objective BLP model for the John Wiley and Sons.
road diet NDP by minimizing the total travel time of both Boyce, D. E. (1984). “Urban transportation network-equilibrium and
cyclists and motorists, or by minimizing the automobile design models: Recent achievements and future prospects.”
share and the average travel time per unit distance for motor- Environment and Planning A, Vol. 16, No. 11, pp. 1445-1474, DOI:
10.1068/a161445.
ists who keep using automobiles after the implementation of
Brundtland Commission (1987). Report of the world commission on
a road diet. Mesbah et al. (2012) proposed a bi-level optimi- environment and development (http://www.un.org/documents/ga/
zation to design a network of bike lines, where the upper res/42/ares42-187.htm). United Nations.
level considered the benefits of cyclists (total travel distance Chen, A. and Yang, C. (2004). “Stochastic transportation network design
on bike lanes) and car users (total travel time by car), and the problem with spatial equity constraint.” Transportation Research

Vol. 20, No. 4 / April 2016 − 1095 −


Xiangdong Xu, Anthony Chen, and Chao Yang

Record, Vol. 1882, pp. 97-104, DOI: 10.3141/1882-12. Research Record, Vol. 2089, pp. 58-65, DOI: 10.3141/2089-08.
Chen, A., Chootinan, P., and Wong, S. C. (2006). “New reserve capacity Environmental Protection Agency (2008). <http://www.epa.gov/air/
model of a signal-controlled road network.” Transportation Research emissions/index.htm>.
Record, Vol. 1964, pp. 35-41, DOI: 10.3141/1964-05. Farahani, R. Z., Miandoabchi, E., Szeto, W. Y., and Rashidi, H. (2013).
Chen, A. and Kasikitwiwat, P. (2011). “Modeling network capacity “A review of urban transportation network design problems.”
flexibility of transportation networks.” Transportation Research European Journal of Operational Research, Vol. 229, No. 2, pp.
Part A, Vol. 45, No. 2, pp. 105-117, DOI: 10.1016/j.tra.2010.11.003. 281-302, DOI: 10.1016/j.ejor.2013.01.001.
Chen, A., Kasikitwiwat, P., and Yang, C. (2013). “Alternate capacity Feng, T. and Timmermans, H. J. P. (2014). “Trade-offs between mobility
reliability measures for transportation networks.” Journal of Advanced and equity maximization under environmental capacity constraints: A
Transportation, Vol. 47, No. 1, pp. 79-104, DOI: 10.1002/atr.216. case study of an integrated multi-objective model.” Transportation
Chen, A., Kim, J., Lee, S., and Kim, Y. (2010). “Stochastic multi-objective Research Part C, Vol. 43, No. 3, pp. 267-279, DOI: 10.1016/
models for network design problem under demand uncertainty.” Expert j.trc.2014.03.012.
Systems with Applications, Vol. 37, No. 2, pp. 1608-1619, DOI: Feng, T. and Zhang, J. (2014). “Multicriteria evaluation on accessibility-
10.1016/j.eswa.2009.06.048. based transportation equity in road network design problem.”
Chen, A., Kim, J., Zhou, Z., and Chootinan, P. (2007). “Alpha reliable Journal of Advanced Transportation, Vol. 48, No. 6, pp. 526-541,
network design problem.” Transportation Research Record, Vol. DOI: 10.1002/atr.1202.
2029, pp. 49-57, DOI: 10.3141/2029-06. Friesz, T. L. (1985). “Transportation network equilibrium, design and
Chen, A. and Subprasom, K. (2007). “Analysis of regulation and policy aggregation: Key developments and research opportunities.”
of private toll roads in a build-operate-transfer scheme under Transportation Research Part A, Vol. 19, Nos. 5-6, pp. 413-427,
demand uncertainty.” Transportation Research Part A, Vol. 41, No. DOI: 10.1016/0191-2607(85)90041-X.
6, pp. 537-558, DOI: 10.1016/j.tra.2006.11.009. Friesz, T. L., Han, K., Liu, H., and Yao, T. (2013). “Dynamic congestion
Chen, A., Subprasom, K., and Ji, Z. (2006). “A Simulation-based Multi- and tolls with mobile source emission.” Procedia - Social and
objective Genetic Algorithm (SMOGA) for build-operate-transfer Behavioral Sciences, Vol. 80, pp. 818-836, DOI: 10.1016/j.sbspro.
network design problem.” Optimization and Engineering Journal, 2013.05.044.
Vol. 7, No. 3, pp. 225-247, DOI: 10.1007/s11081-006-9970-y. Gao, Z. Y. and Song, Y. F. (2002). “A reserve capacity model of optimal
Chen, A. and Xu, X. (2012). “Goal programming approach to solving signal control with user-equilibrium route choice.” Transportation
network design problem with multiple objectives and demand Research Part B, Vol. 36, No. 4, pp. 313-323, DOI: 10.1016/S0191-
uncertainty.” Expert Systems with Applications, Vol. 39, No. 4, pp. 2615(01)00005-4.
4160-4170, DOI: 10.1016/j.eswa.2011.09.118. Gardner, L. M., Duell, M., and Waller, S. T. (2013). “A framework for
Chen, A., Zhou, Z., Chootinan, P., Ryu, S., Yang, C., and Wong, S. C. evaluating the role of electric vehicles in transportation network
(2011a). “Transport network design problem under uncertainty: A infrastructure under travel demand variability.” Transportation Research
review and new developments.” Transport Reviews, Vol. 31, No. 6, Part A, Vol. 49, pp. 76-90, DOI: 10.1016/j.tra.2013.01.031.
pp. 743-768, DOI: 10.1080/01441647.2011.589539. Hizir, A. E. (2006). Using emission functions in mathematical programming
Chen, A., Zhou, Z., and Ryu, S. (2011b). “Modeling physical and models for sustainable urban transportation: An application in
environmental side constraints in traffic equilibrium problem.” bilevel optimization, Master of Science Dissertation, Sabanci
International Journal of Sustainable Transportation, Vol. 5, No. 3, University, Turkey.
pp. 172-197, DOI: 10.1080/15568318.2010.488277. Ho, H. W. and Sumalee, A. (2010). “Congestion pricing design with
Chen, L. and Yang, H. (2012). “Managing congestion and emissions in environmental justice and utility-based equity.” Proceedings of the
road networks with tolls and rebates.” Transportation Research Part 89th Annual Meeting of Transportation Research Board, Washington
B, Vol. 46, No. 8, pp. 933-948, DOI: 10.1016/j.trb.2012.03.001. D.C., USA.
Chootinan, P. and Chen, A. (2006). “Constraint handling in genetic Kamga, C., Levinson, H. S., and Yazici, M. A. (2012). “Sustainability in
algorithms using a gradient-based repair method.” Computers & transportation: Opportunities and challenges.” Proceedings of the
Operations Research, Vol. 33, No. 8 pp. 2263-2281, DOI: 10.1016/ 91st Annual Meeting of Transportation Research Board, Washington,
j.cor.2005.02.002. D.C., USA.
Chootinan, P., Wong, S. C., and Chen, A. (2005). “A reliability-based Lee, D. H., Wu, L., and Meng, Q. (2006). “Equity based land-use and
network design problem.” Journal of Advanced Transportation, Vol. transportation problem.” Journal of Advanced Transportation, Vol.
39, No. 3, pp. 247-270, DOI: 10.1002/atr.5670390303. 40, No. 1, pp. 75-93, DOI: 10.1002/atr.5670400105.
Connors, R., Sumalee, A., and Watling, D. (2005). “Equitable network Levinson, D. (2010). “Equity effects of road pricing: A review.” Transport
design.” Journal of the Eastern Asia Society for Transportation Reviews, Vol. 30, No. 1, pp. 33-57, DOI: 10.1080/01441640903189304.
Studies, Vol. 6, pp. 1382-1397, DOI: 10.11175/easts.6.1382. Li, Z. C. and Ge, X. (2014). “Traffic signal timing problems with
Current, J. and Marsh, M. (1993). “Multiobjective transportation network environmental and equity considerations.” Journal of Advanced
design and routing problems: Taxonomy and annotation.” European Transportation, Vol. 48, No. 8, pp. 1066-1086, DOI: 10.1002/
Journal of Operational Research, Vol. 65, No. 1, pp. 4-19, DOI: atr.1246.
10.1016/0377-2217(93)90140-I. Li, Z. C., Lam, W. H. K., Wong, S. C., and Sumalee, A. (2012).
Duell, M., Gardner, L., and Waller, S. T. (2013). “Multi-objective traffic “Environmentally sustainable toll design for congested road networks
network design accounting for plug-in electric vehicle energy with uncertain demand.” International Journal of Sustainable
consumption.” Proceedings of the 92nd Annual Meeting of the Transportation, Vol. 6, No. 3, pp. 127-155, DOI: 10.1080/15568318.
Transportation Research Board, Washington, D.C. 2011.570101
Duthie, J. and Waller, S. T. (2008). “Incorporating environmental justice Li, Z. C., Li, Z. K., and Lam, W. H. K. (2014). “An integrated design of
measures into equilibrium-based network design.” Transportation sustainable land use and transportation system with uncertainty in

− 1096 − KSCE Journal of Civil Engineering


A Review of Sustainable Network Design for Road Networks

future population.” Transportmetrica A, Vol. 10, No. 2, pp. 160-185, transportation networks with stochastic dependencies: A theoretical
DOI: 10.1080/18128602.2012.725268. copula-based model.” Networks and Spatial Economics, Vol. 13,
Litman, T. (2014). Developing indicators for sustainable and livable No. 4, pp. 373-397, DOI: 10.1007/s11067-013-9185-7.
transport planning, Victoria Transport Policy Institute. Pronello, C. and Camusso, C. (2012). “A review of transport noise
Liu, H. and Wang, D. Z. W. (2015). “Global optimization method for indicators.” Transport Reviews, Vol. 32, No. 5, pp. 599-628, DOI:
network design problem with stochastic user equilibrium.” 10.1080/01441647.2012.706332.
Transportation Research Part B, Vol. 72, pp. 20-39, DOI: 10.1016/ Pternea, M., Kepaptsoglou, K., and Karlaftis, M. G. (2015). “Sustainable
j.trb.2014.10.009. urban transit network design.” Transportation Research Part A, Vol.
Lo, H. and Szeto, W. Y. (2009). “Time-dependent transport network 77, pp. 276-291, DOI: 10.1016/j.tra.2015.04.024.
design under cost-recovery.” Transportation Research Part B, Vol. Ramjerdi, F. (2006). “Equity measures and their performances in
43, No. 1, pp. 142-158, DOI: 10.1016/j.trb.2008.06.005. transportation.” Transportation Research Record, Vol. 1983, pp. 67-
Lopez, E. and Monzon, A. (2010). “Integration of sustainability issues 74, DOI: 10.3141/1983-10.
in strategic transportation planning: A multi-criteria model for the Rilett, L. R. and Benedek, C. M. (1994). “Traffic assignment under
assessment of transport infrastructure plans.” Computer-Aided Civil environmental and equity objective.” Transportation Research
and Infrastructure Engineering, Vol. 25, No. 6, pp. 440-451, DOI: Record, Vol. 1443, pp. 92-99.
10.1111/j.1467-8667.2010.00652.x. Santos, B. F., Antunes, A. P., and Miller, E. J. (2010). “Interurban road
Luathep, P., Sumalee, A., Lam, W. H. K., Li, Z. C., and Lo, H. K. (2011). network planning model with accessibility and robustness objectives.”
“Global optimization method for mixed transportation network design Transportation Planning and Technology, Vol. 33, No. 3, pp. 297-
problem: A mixed integer linear programming approach.” 313, DOI: 10.1080/03081061003732375.
Transportation Research Part B, Vol. 45, No. 5, pp. 808-827, DOI: Santos, B., Antunes, A., and Miller, E. (2008). “Integrating equity
10.1016/j.trb.2011.02.002. objectives in a road network design model.” Transportation Research
Ma, R., Ban, J., and Szeto, W. Y. (2015). “Emission modeling and pricing Record, Vol. 2089, pp. 35-42, DOI: 10.3141/2089-05.
in dynamic traffic networks.” Transportation Research Procedia, Santos, B., Antunes, A., and Miller, E. (2009). “A multiobjective approach
Vol. 9, pp. 106-129, DOI: 10.1016/j.trpro.2015.07.007. to long-term interurban multilevel road network planning.” Journal
Magnanti, T. L. and Wong, R. T. (1984). “Network design and transportation of Transportation Engineering, Vol. 135, No. 9, pp. 640-649, DOI:
planning: Models and algorithms.” Transportation Science, Vol. 18, 10.1061/(ASCE)TE.1943-5436.0000043.
No. 1, pp. 1-55, DOI: 10.1287/trsc.18.1.1. Schweitzer, L. and Valenzuela, A. (2004). “Environmental Injustice and
Maheshwari, P., Kachroo, P., Paz, A., and Khaddar, R. (2015). Transportation: The Claims and the Evidence.” Journal of
“Development of control models for the planning of sustainable Planning Literature, Vol. 18, No. 4, pp. 383-398, DOI: 10.1177/
transportation systems.” Transportation Research Part C, Vol. 55, 0885412204262958.
pp. 474-485, DOI: 10.1016/j.trc.2015.03.024. Sharma, S. and Mathew, T. V. (2011). “Multiobjective network design
Maruyama, T. and Sumalee, A. (2007). “Efficiency and equity comparisons for emission and travel-time trade-off for a sustainable large urban
of cordon- and area-based road pricing schemes using a trip-chain transportation network.” Environment and Planning B: Planning
equilibrium model.” Transportation Research Part A, Vol. 41, No. 7, and Design, Vol. 38, No. 3, pp. 520-538, DOI: 10.1068/b37018.
pp. 655-671, DOI: 10.1016/j.tra.2006.06.002. Sharma, S. and Mishra, S. (2013). “Intelligent transportation systems-
Meng, Q. and Yang, H. (2002). “Benefit distribution and equity in road enabled optimal emission pricing models for reducing carbon
network design.” Transportation Research Part B, Vol. 36, No. 1, footprints in a bimodal network.” Journal of Intelligent Transportation
pp. 19-35, DOI: 10.1016/S0191-2615(00)00036-9. Systems, Vol. 17, No. 1, pp. 54-64, DOI: 10.1080/15472450.2012.
Mesbah, M., Thompson, R., and Moridpour, S. (2012). “Bi-level 708618.
optimization approach to design of network of bike lanes.” Sohn, K. (2011). “Multi-objective optimization of a road diet network
Transportation Research Record, Vol. 2284, pp. 21-28, DOI: design.” Transportation Research Part A, Vol. 45, No. 6, pp. 499-
10.3141/2284-03. 511, DOI: 10.1016/j.tra.2011.03.005.
Miandoabchi, E., Daneshzand, F., Farahani, R. Z., and Szeto, W. Y. Steininger, K. W., Friedl, B., and Gebetsroither, B. (2007). “Sustainability
(2015). “Time-dependent discrete road network design with both impacts of car road pricing: A computable general equilibrium
tactical and strategic decisions.” Journal of the Operational Research analysis for Austria.” Ecological Economics, Vol. 63, No. 1, pp. 59-
Society, Vol. 66, No. 6, pp. 894-913, DOI: 10.1057/jors.2014.55. 69, DOI: 10.1016/j.ecolecon.2006.09.021.
Nagurney, A. (2000). “Alternative pollution permit systems for Subprasom, K. and Chen, A. (2007). “Effects of regulation on highway
transportation networks based on origin-destination pairs and paths.” pricing and capacity choice of a build-operate-transfer scheme.”
Transportation Research Part D, Vol. 5, No. 1, pp. 37-58, DOI: ASCE Journal of Construction Engineering and Management, Vol.
10.1016/S1361-9209(99)00023-1. 133, No. 1, pp. 64-71, DOI: 10.1061/(ASCE)0733-9364(2007)
Nagurney, A., Qiang, Q., and Nagurney, L. S. (2010). “Environmental 133:1(64).
impact assessment of transportation networks with degradable links Sugawara, S. and Niemeier, D. A. (2002). “How much can vehicle
in an era of climate change.” International Journal of Sustainable emissions be reduced? Exploratory analysis of an upper boundary
Transportation, Vol. 4, No. 3, pp. 154-171, DOI: 10.1080/ using an emission-optimized trip assign.” Transportation Research
15568310802627328. Record, Vol. 1815, pp. 29-37, DOI: 10.3141/1815-04.
Nagurney, A., Ramanujam, P., and Dhanda, K. K. (1998).” A multimodal Sumalee, A., Shepherd, S., and May, A. (2009). “Road user charging
traffic network equilibrium model with emission pollution permits: design: Dealing with multi-objectives and constraints.” Transportation,
Compliance versus noncompliance.” Transportation Research Part Vol. 36, No. 2, pp. 167-186, DOI: 10.1007/s11116-009-9197-9.
D, Vol. 3, No. 5, pp. 349-374, DOI: 10.1016/S1361-9209(98)00016-9. Szeto, W. Y., Jaber, X., and Wong, S. C. (2012). “Road network
Ng, M. W. and Lo, H. K. (2013). “Regional air quality conformity in equilibrium approaches to environmental sustainability.” Transport

Vol. 20, No. 4 / April 2016 − 1097 −


Xiangdong Xu, Anthony Chen, and Chao Yang

Reviews, Vol. 32, No. 4, pp. 491-518, DOI: 10.1080/01441647. problem with fuzzy goals.” Transportation Research Record, Vol.
2012.690000. 2399, pp. 23-33, DOI: 10.3141/2399-03.
Szeto, W. Y., Jiang, Y., Wang, D. Z. W., and Sumalee, A. (2015). “A Xu, X., Chen, A., and Cheng, L. (2015). “Reformulating environmentally
sustainable road network design problem with land use transportation constrained traffic equilibrium via a smooth gap function.”
interaction over time.” Networks and Spatial Economics, Vol. 15, International Journal of Sustainable Transportation, Vol. 9, No. 6,
No. 3, pp. 791-822, DOI: 10.1007/s11067-013-9191-9. pp. 419-430, DOI: 10.1080/15568318.2013.777261.
Szeto, W. Y. and Lo, H. K. (2006). “Transportation network improvement Yang, H. and Bell, M. G. H. (1998). “Models and algorithms for road
and tolling strategies: The issue of intergeneration equity.” network design: A review and some new developments.”
Transportation Research Part A, Vol. 40, No. 3, pp. 227-243, DOI: Transport Reviews, Vol. 18, No. 3, pp. 257-278, DOI: 10.1080/
10.1016/j.tra.2005.06.004. 01441649808717016.
Szeto, W. Y., Wang, Y., and Wong, S. C. (2014). “The chemical reaction Yang, H., Bell, M. G. H., and Meng, Q. (2000). “Modeling the capacity
optimization approach to solving the environmentally sustainable and level of service of urban transportation networks.” Transportation
network design problem.” Computer-Aided Civil and Infrastructure Research Part B, Vol. 34, No. 4, pp. 255-275, DOI: 10.1016/S0191-
Engineering, Vol. 29, No. 2, pp. 140-158, DOI: 10.1111/mice.12033. 2615(99)00024-7.
Wang, D. Z. W. and Lo, H. K. (2010). “Global optimum of the linearized Yang, H. and Huang, H. J. (2005). Mathematical and economic theory
network design problem with equilibrium flows.” Transportation of road pricing, Elsevier, Amsterdam.
Research Part B, Vol. 44, No. 4, pp. 482-492, DOI: 10.1016/ Yang, H. and Wang, J. Y. T. (2002). “Travel time minimization versus
j.trb.2009.10.003. reserve capacity maximization in the network design problem.”
Wang, G., Gao, Z., Xu, M., and Sun, H. (2014b). “Models and a Transportation Research Record, Vol. 1783, pp. 17-26, DOI:
relaxation algorithm for continuous network design problem with a 10.3141/1783-03.
tradable credit scheme and equity constraints.” Computers & Operations Yang, H. and Zhang, X. (2002). “Multiclass network toll design
Research, Vol. 41, pp. 252-261, DOI: 10.1016/j.cor.2012.11.010. problem with social and spatial equity constraints.” Journal of
Wang, H., Lam, W. H. K., Zhang, X., and Shao, H. (2015). “Sustainable Transportation Engineering, Vol. 128, No. 5, pp. 420-428, DOI:
transportation network design with stochastic demands and chance 10.1061/(ASCE)0733-947X(2002)128:5(420).
constraints.” International Journal of Sustainable Transportation, Yim, K., Wong, S. C., Chen, A., Wong, C. K., and Lam, W. H. K.
Vol. 9, No. 2, pp. 126-144, DOI: 10.1080/15568318.2012.740147. (2011). “A reliability-based land use and transportation optimization
Wang, J. Y. T., Ehrgott, M., Dirks, K. N., and Gupta, A. (2014a). “A model.” Transportation Research Part C, Vol. 19, No. 2, pp. 351-
bilevel multi-objective road pricing model for economic, environmental 362, DOI: 10.1016/j.trc.2010.05.019.
and health sustainability.” Transportation Research Procedia, Vol. 3, Yin, Y. and Lawphongpanich, S. (2006). “Internalizing emission externality
pp. 393-402, DOI: 10.1016/j.trpro.2014.10.020. on road networks.” Transportation Research Part D, Vol. 11, No. 4,
Wismans, L., van Berkum, E. C., and Bliemer, M. C. J. (2011). pp. 292-301, DOI: 10.1016/j.trd.2006.05.003.
Comparison of multiobjective evolutionary algorithms for optimization Yin, Y. and Lawphongpanich, S. (2007). “Estimating highway investment
of externalities by using dynamic traffic management measures.” requirements with uncertain demands.” Transportation Research
Transportation Research Record, Vol. 2263, pp. 163-173, DOI: Record, Vol. 1993, pp. 16-22, DOI: 10.3141/1993-03.
10.3141/2263-18. Yin, Y., Lawphongpanich, S., and Lou, Y. (2008). “Estimating investment
Wong, S. C. and Yang, H. (1997). “Reserve capacity of a signal- requirement for maintaining and improving highway systems.”
controlled road network.” Transportation Research Part B, Vol. 31, Transportation Research Part C, Vol. 16, No. 2, pp. 199-211, DOI:
No. 5, pp. 397-402, DOI: 10.1016/S0191-2615(97)00002-7. 10.1016/j.trc.2007.07.004.
Wu, D., Yin, Y., Lawphongpanich, S., and Yang, H. (2012). “Design of Yin, Y., Li, Z. C., Lam, W. H. K., and Choi, K. (2014). “Sustainable toll
more equitable congestion pricing and tradable credit schemes for pricing and capacity investment in a congested road network: A goal
multimodal transportation networks.” Transportation Research Part programming approach.” Journal of Transportation Engineering,
B, Vol. 46, No. 9, pp. 1273-1287, DOI: 10.1016/j.trb.2012.05.004. Vol. 140, No. 12, 04014062, DOI: 10.1061/(ASCE)TE.1943-
Xu, X., Chen, A., and Cheng, L. (2013). “Stochastic network design 5436.0000720.

− 1098 − KSCE Journal of Civil Engineering

You might also like