close

Вход

Забыли?

вход по аккаунту

?

%28ASCE%29CO.1943-7862.0001402

код для вставкиСкачать
Multiobjective Framework for Managing Municipal
Integrated Infrastructure
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Soliman Abu Samra 1; Mahmoud Ahmed, Ph.D. 2; Amin Hammad, Ph.D., M.ASCE 3;
and Tarek Zayed, Ph.D., F.ASCE 4
Abstract: The massive number of infrastructure intervention activities occurring in cities leads to detrimental social, environmental,
and economic impacts on the community. Thus, integrating the asset intervention activities is required to minimize the community disruption and maintain an acceptable level of service throughout the assets’ lifecycle. This paper presents an integrated multiobjective asset
management system for the road and water infrastructure that enables asset managers to trade off intervention alternatives and compare
the outcomes of both conventional and integrated asset management systems. The multiobjective framework considers (1) physical state,
(2) lifecycle costs, (3) user costs, and (4) replacement value. It revolves through three core models: (1) a database model that contains
detailed asset inventory for the road and water networks; (2) key performance indicator (KPI) computational models for assessing the impact
of the intervention plan on the predefined set of KPIs; and (3) an optimization model that relies on a combination of metaheuristics, dynamic
programming, and goal optimization to schedule the intervention activities throughout the planning horizon. The system is applied to
road and water networks in Kelowna, British Columbia, and the results show 33 and 50% savings in lifecycle costs and user costs,
respectively. Moreover, it shows the potential ability to scale the framework to include other infrastructure such as sewer, electricity,
gas, and telecom, provided that the information can be shared among these entities. DOI: 10.1061/(ASCE)CO.1943-7862.0001402.
© 2017 American Society of Civil Engineers.
Author keywords: Integrated asset management; Infrastructure planning; Multiobjective optimization; Key performance indicators;
Decision making; Cost and schedule.
Introduction
Infrastructure is the foundation of our daily lives that enables the
communities to prosper and local businesses to grow. Infrastructure
development is a vital component in encouraging a country’s economic growth. Finance Canada recently showed that a $1-billion
investment in infrastructure creates 16,700 jobs and boosts the gross
domestic product (GDP) by $1.6 billion (Wood 2015). Developing
infrastructure enhances a country’s productivity, consequently
making firms more competitive and boosting a region’s economy.
Not only does infrastructure enhance the efficiency of production,
transportation, and communication, it also helps provide economic
incentives to public sector and private sector participants.
However, municipalities are experiencing high inefficiency and
financial burdens imposed by their underperforming infrastructure.
1
Ph.D. Candidate and Graduate Research Assistant, Dept. of Building,
Civil and Environmental Engineering, Concordia Univ., Montreal, QC,
Canada H3G 1M8 (corresponding author). E-mail: solimanamr_8@
aucegypt.edu
2
Ph.D. Candidate and Graduate Research Assistant, Dept. of Building,
Civil and Environmental Engineering, Concordia Univ., Montreal, QC,
Canada H3G 1M8. E-mail: mahm_ah@encs.concordia.ca
3
Professor, Concordia Institute for Information Systems Engineering,
Concordia Univ., Montreal, QC, Canada H3G 1M8. E-mail: hammad@
ciise.concordia.ca
4
Professor, Dept. of Building, Civil and Environmental Engineering,
Concordia Univ., Montreal, QC, Canada H3G 1M8. E-mail: tarek.zayed@
concordia.ca
Note. This manuscript was submitted on January 18, 2017; approved on
June 8, 2017; published online on October 26, 2017. Discussion
period open until March 26, 2018; separate discussions must be submitted
for individual papers. This paper is part of the Journal of Construction
Engineering and Management, © ASCE, ISSN 0733-9364.
© ASCE
One-third of Canada’s municipal infrastructure is in fair, poor, and
failing condition states (FCM 2016). Consequently, the risk of
service disruption dramatically increases, which forces decision
makers to take immediate corrective action to maintain them. Furthermore, aging infrastructure systems worldwide are placing tremendous pressure on governments through steeply growing budget
deficits and urgent need for replacement. For example, Canada’s
municipal infrastructure deficit was estimated at $123 billion for
the existing infrastructure, growing by $2 billion annually, with
$115 billion needed for constructing new infrastructure (Mirza
2009) to satisfy the growing population, which increased from
17.9 million in 1960 to 35.1 million in 2013 and is expected to
be 42.5 million by 2056 (Statistics Canada 2015). According to
the United Nation Population Division, the world is undergoing the
largest wave of urban growth; in 2008, more than 50% of the
world’s population lived in towns and cities, and the figure is expected to exponentially swell throughout the upcoming years
(Osman 2015). The need for asset management (AM) adoption
has been strengthened by the plethora of infrastructure problems
such as deteriorating condition, decreasing level of service
(LOS), aging infrastructure, and urbanization. Meanwhile, municipalities are utilizing a conventional AM approach in which the
intervention actions for spatially located assets are undertaken separately on an asset level, without accounting for the potential
integration that might be a feasible alternative in some cases.
However, an integrated AM approach enables asset managers to
integrate the intervention actions for assets with the capability
of undertaking separate asset-based interventions whenever necessary. This in turn enables asset managers to account for the assets’
interdependencies (i.e., physical, cyber, geographical, and logical)
in the decision-making process, which results in less service
disruption along with time, monetary, and spatial savings due to
04017091-1
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
the lesser number of interventions required for maintaining all the
assets in the street section.
Therefore this paper develops a holistic AM approach that
integrates the intervention actions carried out for both roads and
water networks. Accordingly, the objectives are as follows:
1. Establish the key performance indicator (KPI) selection criteria
for corridor infrastructure assets;
2. Build an integrated AM system for roads and water networks;
3. Develop a comparison approach to calculate the benefits of the
integrated AM approach over the conventional AM approach;
and
4. Optimize the selection of the asset intervention plans for both
the conventional and integrated AM systems.
Background
Throughout the last decade, many researchers have studied the
concept of integrating the planning and funding of two or more
networks of colocated assets in the right-of-way corridor (Halfawy
2008, 2010; Halfawy and Figueroa 2006; Infraguide 2003a, b, c,
2004, 2006). The development of integration approaches and AM
tools is still evolving. The mainstream of integrated AM research
has concentrated on a number of perspectives, such as system complexity, where it enhances the management and balance of the
factors from both micro/project-level and macro/network-level perspectives (Houston 2014); financial reporting, where the goal is
reaching a near-optimum funding decision through implementing
cost-benefit analysis to undertake various intervention alternatives
(Halfawy 2008); information management, where the integrated
AM represents an excellent example of Big Data, which requires
improving the data integration process across multiple departments
(Disivo and Ladiana 2011); integration factors, which are restricted
to specific areas such as geographic information system (GIS)based integration systems (Halfawy 2010), integrated risk-based
decision making (Shahata and Zayed 2010), and integrated condition rating (Elsawah et al. 2014); and conflicting perspectives, because integrated AM could be taken from both a community-based
perspective and a municipal perspective with changing objectives
(Khan et al. 2015).
Few municipalities have an integrated AM plan for their road,
sewer, and water systems (InfraGuide 2003a, 2006). Although
many municipalities have implemented pavement management systems, most municipalities do not have AM plans for their water and
sewer systems due to the longer service life and more complicated
condition assessment of these systems. Few scholars have gone
beyond presenting the concept and published dedicated research
on the integration and coordination of municipal maintenance
and rehabilitation planning. Abu-Samra et al. (2017) developed
AM decision-making optimization systems for scheduling the intervention activities for multiple highways using goal optimization for
a predefined set of KPIs. Other scholars developed coordination
frameworks using simulation of both the spatial and temporal
dimensions with the objective of minimizing the lifecycle costs
(LCCs) (Oh et al. 2011; Amador and Magnuson 2011). Coordination of intervention activities has been thoroughly considered
as a part of the wider notion of the dependency and interdependency
relationships between infrastructure systems. The dependency between the systems refers to a unidirectional relationship in which
one system relies on the other, whereas interdependency refers
to the bidirectional relationship between the infrastructure systems
(Rinaldi et al. 2001). It could be classified as functional and spatial
depending on the operational dependency and the proximity between systems, respectively (Zimmerman 2010). It also can be
© ASCE
classified based on the interdependency category, i.e., physical,
cyber, geographical, and logical (Rinaldi et al. 2001). Multiple computational approaches have been used, such as simulation, econometrics, network-based analysis, and system-of-system modeling.
Ouyang (2014) conducted extensive state-of-the-art reviews to summarize this work. Despite the fact that plentiful modeling computational approaches have been utilized in the last decade, some
limitations have been noticed: (1) the propagation of the system disruption has not been appropriately considered, because the vast
majority of the research focused on the operation phase; (2) the dimension of time was not considered as a key aspect that influences
the AM intervention decisions; and (3) there has been a lack of focus
on holistic-based intervention for interdependent colocated infrastructure systems (i.e., roads, water, and sewer).
As a part of the holistic AM decision-making systems, several
scholars developed integrated AM multiobjective decision-making
frameworks to analyze the trade-off between delaying and bringing
forward the intervention activities for road, water, and sewer
networks. The integration dimensions varied from one scholar to
another; some scholars relied on a single objective, i.e., risk or
LCC, when scheduling their intervention action plans throughout
the planning horizon. Scholars accounted for both the probabilities
and consequences of failures when developing an amalgamated
risk index (RI) using a mixed Delphi-analytical hierarchal process
(Delphi-AHP) approach, and RI through K-means clustering
(Shahata and Zayed 2016; Carey and Lueke 2013). However,
other scholars relied on the multiobjective nature of the problem,
i.e., financial (LCC), physical/serviceability (LOS), and risk, and
utilized multiobjective optimization techniques to schedule their
intervention activities while minimizing the overall objectives’
deviations (Osman 2015). Similarly, other scholars framed an integrated strategic AM approach that prioritizes the street sections that
need rehabilitation by considering the road network and its underlying infrastructure, i.e., sewer and water supply networks. In addition, Tscheikner-Gratl et al. (2015) developed a priority model (PM)
that includes the pipes’ deterioration in terms of discharge water
in water supply networks and urban flooding in sewer networks.
The domain of infrastructure KPI definition and selection
has not received enough attention from researchers. However, a
few researchers and several guidelines, e.g., ISO 55000 (ISO 2014)
standards for asset managers, and municipalities’ annual reports,
have established criteria for selecting the KPIs (Infraguide
2003a, b, c, 2004, 2006; Khan et al. 2015), Most of them advocated
the transition from a condition-centric and cost-centric asset stewardship approach to an asset serviceability approach that considers
cost, physical state, criticality, and risk exposure (NAMS 2015;
Osman 2015). In addition, some guidelines focused on setting specific, measurable, achievable, realistic, time-based (SMART) rules
that need to be met prior to selecting any KPI (Table 1). Other
scholars defined sets of KPIs for assessing the performance of the
infrastructure, including LOS effectiveness, cost efficiency, timeliness of response, safety procedures, and quality of service from the
agency and user satisfaction points of view (Abu-Samra et al. 2017;
Pinero and de la Garza 2003).
The application of multiobjective optimization within the
domain of infrastructure AM has received considerable attention
from researchers. Rashedi and Hegazy (2014) compared segmented
genetic algorithms and exact numerical optimization methods
(GAMS/CPLEX) to the capital renewal planning of large infrastructure systems and concluded that numerical methods are superior.
Furthermore, numerous research has investigated multiobjective
algorithms, including linear programming and integer programming, and implemented them in various applications, including
network-level pavement intervention action plans and large-scale
04017091-2
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Table 1. Integrated KPI Rules
Indicator category
Accuracy
Frequency
Financial
Ownership
Portability
Subjectivity
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Understandability
Indicative rules
The indicator should consider the precision level needed while measuring
The indicator should consider the difficulty level, represented by either the frequency or easiness of measurement
The indicator should frequently (i.e., annually) track the asset’s performance throughout the planning horizon
The indicator’s rate of change should be highly considered (i.e., the KPI should experience a periodical difference in the asset state)
The indicator should consider the costs needed for frequently measuring and controlling the asset
The indicator should have an owner who is held liable/responsible for it
The indicator should fit multiassets with different features and attributes such as deterioration rates, useful lives, and construction years)
The indicator should be objective and should include a predefined set of rules for measuring an asset attribute to guarantee a consensus
agreement among different parties
The indicator should consider the ease of understanding and tracing the triggers behind a sudden rise/fall throughout the asset’s lifecycle
building component repair (De la Garza et al. 2011; Hegazy and
Elhakeem 2011; Tong et al. 2009).
network-based, to align it with the overall municipality goals, represented through KPI thresholds.
Conventional AM System
Developed Approach
The framework supports the transition from a condition-centric
and cost-centric asset stewardship approach to an asset serviceability approach that considers cost, physical state, criticality, and risk
exposure. The developed approach functions through two different
systems: (1) a conventional AM system, and (2) an integrated AM
system. The main objective is to compare both systems, visualize
the benefits of integration over the conventional AM, and show the
savings/enhancements in terms of KPIs.
The presented AM system revolves through seven integrated
models (Fig. 1). The database model contains a detailed asset inventory for the roads and water networks. In addition, the KPI definition model outlines the criteria for selecting the most appropriate
KPIs for the assets under study. Afterward, in the core of the
system, a future prediction and deterioration model features three
different deterioration patterns: (1) Markov-based deterioration;
(2) regression-based deterioration; and (3) Weibull-based deterioration. The Markov-based deterioration was developed to feed the
regression-based deterioration with a pattern that reflects the annual
KPIs’ decrease in terms of age for the roads network. However, the
Weibull-based deterioration model was developed to predict the
physical deterioration of the water networks. Knowing the service
life for each pipe material, the diameter, and the current age, the
Weibull model uses preassumed shape and scale factors that
represent the deterioration rate and characteristic life of the asset
to predict the probable deterioration model throughout the lifecycle
(Kimutai et al. 2015).
The financial part takes place through three computational models: (1) an economic/financial model represented by the LCC, the
analysis of which is carried out to estimate the LCC throughout
the planning horizon; (2) a physical/financial model represented
by the replacement value, which, similar to the economic loss,
is the estimated monetary value of the assets given their physical
condition state; and (3) a social/financial model represented by the
user costs, which accounts for the indirect user costs incurred due to
the systems’ failure or asset maintenance. Finally, because of the
assets’ different physical states through the remaining service life,
the optimization model formulates the problem and summarizes the
attributes, including the objective function, variables, and constraints, in an organized and user-friendly interface for the system
users to easily pick the optimization attributes and run the engine
accordingly. In addition, the system features a plot of KPI results
that gives the user the opportunity to track the intervention scenarios’ influence on higher management levels, i.e., area-based and
© ASCE
Database Model
The database model consists of four tables: (1) an asset inventory,
which represents all the attributes, including physical, and spatial
attributes, of the municipal assets under study; (2) an intervention
actions table, which lists the intervention activities of each asset
along with their time and cost implications and heir effects on
the asset service life; (3) a KPI table, which lists the KPIs along
with their categories, units of measurement, and thresholds; and
(4) a user costs table, which lists the truck versus passenger car
distribution percentage along with the estimated user cost per
minute delay for each area.
KPI Definition Model
There is a need for careful KPI selection because of the multiasset
nature of the model with different characteristics such as deterioration rates, useful lives, installation years, intervention costs, and
physical condition implications. Thus there should be common
comparison criteria to assess the different assets’ performance
throughout the planning horizon. Moreover, the established KPIs
need to be indicative, specific, measurable, achievable, realistic,
and timely to predict their annual performance, from economic, financial, physical, and social perspectives, before and after applying
different intervention plans. Accordingly, after conducting an exhaustive literature review, a set of categories and rules was selected
to define the KPIs (Table 1). Based on those rules, four KPIs were
chosen: (1) physical state (PS), (2) replacement value (RV),
(3) LCC, and (4) user costs (UC) (Table 2). The PS is obtained
from the asset deterioration model along with the cost of expected
intervention activities throughout the asset lifecycle. It represents
the condition of the asset under study and is given as a percentage,
where 0% represents an asset in failing condition and 100% represents an asset in excellent condition. The LCC can be obtained
from a financial analysis that considers the impact of decreasing
ownership costs and increasing operating and maintenance costs
of the asset. The UC can be obtained by estimating the indirect user
costs associated with system failure or asset maintenance (Qin
and Cutler 2013). The RV can be obtained by integrating the financial analysis and the assets’ deterioration to reflect the physical/
financial strength of the existing infrastructure.
Future Prediction and Deterioration Model
The system developed three types of deterioration prediction models: (1) a Markov-chain deterioration model composed of a fivecondition transition probability matrix that represents the transition
probabilities of the road sections from one condition state to the
04017091-3
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Fig. 1. Integrated asset-management framework
Table 2. Integrated KPI Definitions
KPI
Physical state (PS)
Lifecycle costs (LCC)
Replacement
value (RV)
User costs (UC)
© ASCE
Description
Category
Unit of
measurement
Represents the condition of the asset under study
Represents total operation and maintenance costs throughout the life cycle time
Represents the estimated monetary value of the economic loss for the asset
deterioration at a certain time
Represents the indirect costs that users will pay due to the systems’ failure or the
assets’ maintenance
Physical indicator
Economic/financial indicator
Physical/financial indicator
Percentage
Monetary ($)
Monetary ($)
Social/economic indicator
Monetary ($)
04017091-4
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
lower state; (2) a regression-based deterioration model which
models the road sections’ annual deterioration and represents their
physical state (PSj ); the regression-based deterioration equation
was extracted from the Markov-based model as a function of
the asset age; and (3) a Weibull-based deterioration model which
models the deterioration of the water pipes and calculates the KPIs’
performance throughout the planning horizon.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Lifecycle Costing Model
The lifecycle cost model calculates the LCC of each intervention
scenario by implementing the concepts of time value of money
(Abu-Samra et al. 2017).
User Costs Model
The user costs are “the estimated daily cost to the traveling public
resulting from the construction work being performed” (Daniels
et al. 1999). They are also “the consideration of opportunity cost
of time for drivers when inconvenienced due to infrastructure overtime (Matthews 2003).” They refer to the lost time caused by several factors including detours and rerouting that add to travel time,
reduced roadway capacity that slows the travel speed and increases
the travel time, and delays of opening new or improved facilities
that prevent users from gaining travel-time benefits. Accordingly,
they were calculated based on the following factors:
1. Average annual daily traffic (AADT) for each section,
2. User costs per vehicle per hour ($=vehicle=hour),
3. Passenger cars versus trucks percentage (%),
4. Speed limit (km=hour), and
5. Intervention time per unit length (hour=km).
The user cost model extracts the unit user cost and the cars’
distribution from the user cost database model. Then it extracts
the intervention time per unit length from the intervention actions
database. Finally, the section travel time is calculated, based on
both the road average speed and the road length (Qin and
Cutler 2013; TDOT 2016). This process is repeated for each optimization engine scenario because the intervention actions differ
from one plan to another
TUCj ¼
LengthðKmÞ
V
½ð1 − T PER Þ AADT PUC Total
SpeedðKm
hr Þ 60
þ ðT Per AADT T UC Þ
ð1Þ
TUCi ¼
m
X
TUCj
ð2Þ
TUCi
ð3Þ
j
TUCT ¼
n
X
i
where V Total = number of vehicles passing per section; T Per = truck
percentage per section (%); PUC = passenger cars unit cost ($);
T UC = trucks unit cost ($); j = section counter; m = total number
of sections; TUCj = total daily section user costs considering the
travel time ($/day); TUCi = annual user costs for all m sections ($);
i = age counter; n = planning horizon; and TUCT = cumulative user
costs throughout the planning horizon n ($).
Replacement Value Model
The replacement value is estimated as the dollar value of replacing
the asset at a certain time, based on the deteriorated performance. In
other words, it reflects the economic loss at the time of replacement, not the full cost of replacement [Eq. (4)]. However, the full
© ASCE
replacement value can be easily estimated using the same approach
by assuming an asset at its failure condition (PS ¼ 0). Thus Eq. (4)
computes the section replacement value (RV j ), which is a part of
the section replacement cost (RCj ), depending on the asset’s condition state. Eq. (5) sums the section replacement value (RV j ) to
calculate the annual replacement value (RV i ) to indicate the total
economic loss in a certain year (i). Finally, Eq. (6) computes the
cumulative replacement value (RV T ) to indicate the cumulative
economic loss until the end of the planning horizon (n)
RV j ¼ ð1 − PSj Þ RCj ð1 þ inÞn
RV i ¼
m
X
ð4Þ
RV j
ð5Þ
RV i
ð6Þ
j
RV T ¼
n
X
i
where RV j = section’s replacement value based on its current
physical state ($); PSj = section physical state (%); RCj = total
replacement cost of a section ($); in = annual inflation rate (%);
RV i = annual replacement value for all m sections ($); and
RV T = cumulative replacement value throughout the planning
horizon n ($).
Optimization Model
The need for optimization was obvious because of the numerous
possible and valid solutions for intervention plans. For instance,
the number of solutions for 20 sections in a water network, through
a 25-year planning horizon, incorporating five intervention actions
is 520×25 . In addition, limited budgets and increasing demand for
higher LOS place extra constraints on asset managers to optimally
choose intervention plans that fulfill the end users’ expectations.
Moreover, the extensible planning horizon, i.e., 25 years, makes
it even more computationally complicated and challenging, creating extra uncertainty for asset managers when making interventionrelated decisions.
Accordingly, this was the main motive behind using a genetic
algorithm (GA) engine, combined with preapplied metaheuristic
rules that minimize the optimization engine search space. Genetic
algorithms are derived from biological systems—they rely on simulating the natural survival of the fittest, where the solution is
represented as a string of chromosomes, which consist of several
genes. The genes’ exchanging process within the chromosomes is
carried out through mutation and crossover operations, and new
solutions are evaluated to replace the weaker members in the population to produce better solutions. This process continues until a
near-optimum solution is generated. The GA performance is affected by four main parameters: (1) number of generations, (2) population size, (3) mutation rate, and (4) crossover rate (Elbeltagi and
Tantawy 2008).
The system uses advanced spreadsheet modeling and Evolver as
an optimization engine. It functions through a powerful engine that
is designed to fit the municipalities’ needs and meet the KPIs’ defined thresholds. The engine applies the following predetermined
metaheuristic rules to lessen the search space for the optimization
engine: (1) the total annual interventions per area should not exceed
two actions, (2) the interdisruption time (IDT) per section between
carrying out one intervention and another should be more than
5 years to minimize the service disruptions, and (3) the total number of interventions per section/group should not exceed five interventions per 25-year planning horizon.
04017091-5
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
After the decision variables are identified, the constraint formulation uses binary coding rules, where 0 represents meeting the
constraint and 1 represents failing to meet the constraint, detailing
the annual status of each KPI. Then the decision variables are formulated through integer programming, in which the decision variables’ ranges are defined according to the number of interventions
that need to be addressed in the model. The wider the range of the
decision variables, the exponentially more computationally complicated the optimization problem is, because it generates a greater
spectrum of possible combinations and requires an evolutionary
optimization algorithm that is able to reach a near-optimum solution for this combinatorial problem. Finally, dynamic programming
is used to account for the extensible planning horizon (n). Five-year
segmentation analysis was chosen for applying dynamic programming and the optimization was subsequently done throughout the
whole planning horizon in a chronological sequence. The optimization attributes are similar for both assets and are divided into segments; the planning horizon is divided into five segments. They
could be mathematically formulated as
Minimize LCCT
ð7Þ
Subject to the following constraints:
LCCi ≤ LCCB
ð8Þ
PSj ≥ PSmin
ð9Þ
TUCi ≤ TUCB
ð10Þ
RV i ≤ RV B
ð11Þ
where LCCT = cumulative lifecycle costs throughout the optimization planning horizon n=5 ($); LCCi = annual lifecycle costs ($);
LCCB = annual budget threshold ($); PSmin = physical state threshold (%); TUCB = annual user costs threshold ($); and RV B = annual
replacement value threshold ($).
Integrated AM System
Asset managers need to utilize a holistic approach in their decisionmaking process rather than the conventional approach when
the integration is feasible/applicable and is both technically and
economically recommendable. Proper coordination of the infrastructure intervention activities will result in (1) minimized community disruption and discomfort; (2) reduced work duplication
occurrence; (3) enhanced intervention efficiency and effectiveness,
through diminishing the intervention time and utilizing better
management and scheduling approaches such as a lane rental approach (Scott et al. 2006); (4) minimized LCC and maximized LOS
across various infrastructure systems rather than making suboptimal decisions for each system separately; (5) much more efficient
tendering and contracting for the construction work; and (6) projects being more attractive for contractors due to the ability to
package construction works in larger projects, especially for small
communities.
Integration Characteristics
The integration between different assets with different characteristics, i.e., useful lives, deteriorating mechanisms, intervention
schedule and techniques, and so on, requires multilevel hierarchal
decision-support systems to integrate the different assets. The bilevel hierarchal integration dimensions are represented as follows:
(1) KPI weights or level of importance, and (2) asset weights or
© ASCE
level of importance. The KPI levels of importance were assumed
to be 50, 10, 15, and 25% for LCC, RV, UC, and PS, respectively.
However, the user can input any other values depending on case
study and priorities and goals of top management, which differ
from one city to another. The asset levels of importance were assumed to be 40% for the road network and 60% for the water
networks, due to their complexity and the additionally incurred
intervention costs.
The integrated AM uses the same conventional databases as
input. However, the deterioration modeling and optimization models differ from the conventional systems due to the difference in
the intervention scenarios, i.e., integrated intervention between
two assets, which share some common activities such as excavation, backfilling, site reinstatement works, setting up traffic control
devices, traffic diversions, notification of residents, and so on. The
difference in the intervention scenarios influences the physical
state, LCC, user costs, and replacement value, which should be
considered in both the deterioration modeling and the optimization
models. The deterioration modeling calculates the effect of the
intervention action on the physical state, which in turn affects
the replacement value. In addition, the LCCs are minimized in
the case of integration due to the removal of common activities
that are duplicated in the conventional management systems.
Furthermore, the user costs are significantly minimized because
the time of disruption in the case of integrated activities is much
less than the time of disruption in the case of conventional-based
interventions. This is due to the removal of both work duplication
activities and interdisruption time, which represents the lag time
between two interventions carried out in the conventional systems for the two assets (Fig. 2). Finally, the regular operation
time (ROT) represents the operation time until the next major
intervention.
To combine the network results for the physical state, the section
length was used to determine the percentage of each area from the
network. The area weights were determined based on the percentage of the section lengths over the total network length. For instance, Area 1 had five sections with a total length of 2.04 km
and the total network length was 8.89 km. Thus, the Area 1 weight
would be computed as the percentage of the total length of sections
in Area 1 divided by the total network length, 22.5% [Eq. (12)].
The same calculation was carried out for the other three areas. Thus
the area weights were 22.5, 30.75, 21, and 25.75% for Areas 1, 2, 3,
and 4, respectively. Accordingly, the physical state of the network
was computed by Eq. (13). All other indicators were summed because they are represented in monetary terms
Lh
Ln
ð12Þ
r Pq
X
o¼1 ðLo PSo Þ
Lh
h¼1
ð13Þ
W areah ¼
PSNet ¼
where Lh = total length of the sections within an area (m); Ln =
total length of the sections within the network; PSNet = physical
state of the network (%); o and h = section and area counters, respectively; q and r = total number of sections and areas, respectively; Lo = length of the section (m); and PSo = physical state
of the section (%).
Integrated AM Optimization
The optimization model differs from the conventional systems
because the integration case is more complex due to the duplicated
number of variables taking place for the two assets that makes
04017091-6
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Fig. 2. Conventional versus integrated system disruption times
it combinatorial in nature and requires powerful optimization engines to reach a near optimal solution. The engine applies a number
of predetermined metaheuristic rules to lessen the search space for
the optimization process as follows: (1) the total annual interventions per area should not exceed two actions, (2) the interdisruption
time between carrying out one intervention and another should be
more than 5 years for each section to minimize the service disruptions, and (3) the total number of interventions per section/group
should not exceed five interventions throughout the planning
horizon.
The decision variable formulation takes place through bidimensional integer programming, in which the first variable’s dimension
ranges are defined according to the type of asset and the number of
available intervention strategies that are considered in the model.
However, the second variable’s dimension are autocalculated by
the model to account for the aforementioned combination challenges with respect to the combined/integrated activities that share
common/duplicated work. Thus the second dimension is represented by the following:
1. 0 represents the do-nothing and the road-intervention scenarios,
in which either no intervention takes place in the period or only
road intervention takes place. Road intervention was combined
with the do-nothing scenario because it does not have anything
to do with the integration. Therefore the disruption caused by
the road will not affect the water network.
2. 1 represents the water intervention–only scenario, in which the
water network is undergoing an intervention and the road is not.
In this case, geographical interdependency occurs because the
two assets are spatially located at the same location and the disruption of the water network will partially/fully affect the road
service. In addition, assuming the absence of trenchless technologies, the intervention of the water network requires excavating
the section, which implies reconstructing the road section and
returning it to a pristine condition state. Accordingly, this case
leaves the contractor with no choice but to implement trenchless
technology for the water-only intervention scenario without
affecting the road.
3. 2 represents the integration scenario, in which the intervention
actions are already integrated. However, it does not make any
sense to carry out slurry seal or crack filling when rehabilitating/
replacing a certain water section. Thus the reconstruction
option takes place in this case to fit the common logic behind
integrating/coordinating the intervention scenarios.
© ASCE
The wider the range of the defined decision variables, the exponentially more complicated the optimization problem is, generating
a greater spectrum of possible combinations and requiring an evolutionary optimization algorithm that can reach a near-optimum
solution for solving such combinatorial in nature problem. Hence
dynamic programming was utilized to overcome the extended
planning horizon issue. Thus 5-year segmentation analysis was
chosen for applying dynamic programming and the optimization
was subsequently done throughout the whole planning horizon
in a chronological sequence. Because goal programming (GP) or
goal optimization was chosen for the integrated management
system, due to conflicting goals and multiassets, there are no constraints because the objective is linked to the variables through goal
constraints. However, the objective is clearly formulated to minimize the sum of deviations from the prescribed goal values defined
by the user. To combine the multiobjectives, a percentile ranking
approach was utilized by computing the percentage deviation
from a goal rather than the absolute deviation (Schniederjans 1995).
Finally, the deviational variables were defined accordingly to
fit the predefined set of KPIs as shown in the aforementioned
equations
m X
n
X
MinðZÞ ¼
þ
was wkl ðd−
i þ di Þ
ð14Þ
j¼1 i¼1
where Z = minimized value of the sum of all negative (d−
i ) and positive (dþ
i ) deviations for n goals; W as = second level of weights
among the assets considered under study (%); W kl = deferential
weights among the conflicting goals (%); di = deviational variables
(%); and i and j = age and asset counters, respectively
d−
1 ¼
dþ
2
d−
3 ¼
04017091-7
J. Constr. Eng. Manage., 2018, 144(1): 04017091
¼
Pn
PSi ðTPSi Þ − PSi ðtÞ
i¼1P
n
i¼1 PSi ðTPSi Þ
Pn
ð15Þ
RV i ðtÞ − RV i ðTRV i Þ
i¼1P
n RV ðTRV Þ
i
i
i¼1
ð16Þ
LCCi ðtÞ − LCCi ðTLCCi Þ
i¼1 P
n LCC ðTLCC Þ
i
i
i¼1
ð17Þ
Pn
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
d−
4
¼
Pn
UCi ðtÞ − UCi ðTUCi Þ
i¼1P
n
i¼1 UCi ðTUCi Þ
ð18Þ
where PS = physical state at a certain time/year (i) for a certain
asset (j) (%); TPS = prescribed goal for the minimal acceptable
physical state per section (%); RV = cumulative replacement value
at time (i) for a certain asset (j) ($); TRV = prescribed goal for the
maximum value amount for network replacement ($); LCC = total
lifecycle costs at a certain time (i) for a certain asset (j) ($); TLCC
= prescribed goal for the maximum monetary lifecycle costs per
network ($); UC = total network user costs at a certain time (i) for
a certain asset (j) ($); and TUC = prescribed goal for the maximum monetary user costs per network ($).
Implementation and Analysis: Kelowna Case Study
To demonstrate the functionality of the coordination framework,
the developed systems were applied to the road and water networks
in Kelowna, located in the southern interior of British Columbia,
Canada. A 9-km stretch of the network was selected for analysis
and divided into four areas. The network includes 20 road sections
covering water pipes of different materials. The date of construction, physical attributes, and condition assessment information
were used to construct the deterioration curves for the physical state
of both road and water networks. For the road network, the pavement condition index (PCI) and international roughness index (IRI)
were used as the baseline condition state. Furthermore, deterministic deterioration curves were used to build the model (Abu-Samra
et al. 2017). The data for the water network were obtained from the
open source City of Kelowna GIS maps (Fig. 3), lacking the condition rating information, which was approximated to mimic the
latest Canadian Infrastructure Report (City of Kelowna 2016;
FCM 2016). Future deterioration prediction was forecast based
on a Weibull-based deterioration model (Kimutai et al. 2015).
Estimated operation and maintenance costs experienced by the road
and water municipalities were used to develop the LCC profile.
Time value of money was considered with an interest rate of
2%. The data set scales/sizes in terms of the number of colocated
road sections and pipes were scaled down several times to enable
the use of the optimization techniques, as highlighted previously.
The study considered the sections where the road and water mains
coexist and have the same length, as shown in Table 3.
Using conventional AM heuristics, the optimal interventions
were selected based on the acceptable thresholds for PSi, RV i ,
LCCi , and UCi . These thresholds were determined based on
(1) meetings with AM experts, (2) previous criticality and consequences of failure modeling (Shahata and Zayed 2016), (3) financial
information about operations and maintenance (Abu-Samra et al.
2017), and (4) the Canada Infrastructure report (FCM 2016).
Because of the importance of the road network, with significant
traffic volumes, the minimal acceptable physical state was 50%
for roads, as opposed to 40% for water networks.
Conventional AM System Results
Road Network Results
The optimal intervention times were determined for the road
network based on the KPI thresholds. The road interventions
were driven by the cost because of the ability to undertake
a cost-effective intervention earlier rather than postponing the
intervention and undertaking costly road rehabilitation. The
conventional AM system for the road network showed reliable
results in terms of meeting the network KPIs. The physical
state of the road network met the threshold along the 25-year
planning horizon because the optimization engine chose to
Fig. 3. Sample from the city of Kelowna GIS maps (infrastructure drawings map viewer) (© 2016 Google, Image City of Kelowna)
© ASCE
04017091-8
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Table 3. Road and Water Networks Database
General characteristics
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Section
identifier
number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Section
length
(m)
370
370
452
393
419
766
451
311
425
783
318
162
498
686
207
715
270
217
519
560
Road network characteristics
Number
of lanes
Section
area
(m2 )
Average
annual daily
traffic (AADT)
Traffic
growth rate
(%)
Current
condition
(%)
Year of
installation
Pipe
diameter
(cm)
Pipe
material
3
4
4
2
3
4
4
2
4
4
3
4
4
4
2
3
2
2
2
4
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3,330
4,440
5,424
2,358
3,771
9,192
5,412
1,866
5,100
9,396
2,862
1,944
5,976
8,232
1,242
6,435
1,620
1,302
3,114
6,720
12,000
8,000
10,000
11,000
7,000
9,500
10,500
8,500
6,800
7,500
9,000
6,000
5,000
11,000
10,000
6,000
9,000
12,000
9,000
8,000
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
90
70
85
65
70
90
70
85
65
70
90
70
85
65
70
90
70
85
65
70
1953
1982
1976
1958
1965
1991
1992
1977
1982
1991
1972
1960
1979
1953
1960
1977
1986
1992
1975
1987
450
150
250
200
150
100
500
500
350
500
500
250
250
100
450
200
150
350
300
150
PVC
PVC
PVC
ASB
ASB
ASB
ASB
ASB
ASB
ASB
ASB
ASB
DI
PVC
ASB
ASB
DI
PVC
ASB
ASB
intervene early for most of the sections, because late interventions
increase the cost by 6–8 times, with a lower physical state.
However, the replacement value increased between 2023 and
2028, which reflects the decreasing physical state and the additional costs required to return the network to a pristine physical
state. Furthermore, the user costs were high in the first year
due to the numerous road interventions carried out. Hence the
modeling was limited to the times when any intervention takes
place to indicate the indirect costs users are paying due to the
closure of a certain section. Finally, the lifecycle costs followed
a similar pattern to the user costs because they reflect the costs
of the interventions. The LCC increased in the first two years
due to the numerous road interventions, but it dramatically decreased over the lifecycle because the optimization chose to carry
out early road interventions, which cost 6–8 times less than late
rehabilitation interventions.
Water Network Results
The optimal intervention times were determined for the water
network based on the KPI thresholds. The water network interventions were driven by the reliability because of the high break rates
that occurred in the water pipes and the subsequent water outages
and repairs that were required. To combine the reliability/physical
state network results, the same rules were used as in Eq. (13). However, the other indicators were summed because they are represented in monetary terms.
The conventional AM system for the water network showed less
promising results in terms of meeting the network KPIs. The reliability of the water network met the threshold along the 25-year
planning horizon. However, from an area perspective the model
was unable to meet the reliability constraint due to the limited annual budget and tried to intervene for the sections with higher
criticality/impact on the overall network in terms of length. Accordingly, the network reliability proved to be stable in meeting the
threshold. The results for the replacement value showed a failure
to meet the reliability threshold. The replacement value barely met
the annual constraint and was constantly high due to the limited
© ASCE
Water network characteristics
Lane
width
(m)
budget for carrying out network repairs. User costs were very
low because the model chose to carry out costly trenchless technology actions. However, realistically, this resulted in an extremely
high lifecycle cost, which exceeded the annual budget in years
2022 and 2037. From an area-level perspective, the model exceeded the annual budget in some areas because the area budget
was distributed on the basis of the section length and the annual
budget was not moved to the next year if it was not properly
utilized.
Integrated AM versus Combined Conventional
AM Results
The goal optimization engine showed promising results for the
integrated AM system in terms of (1) number of interventions,
(2) delay time for service disruptions, (3) user costs resulting from
the delay, (4) combined interventions for the road and water networks, and (5) lifecycle costs throughout the planning horizon. It
met all the KPIs on both a network and an area basis. Figs. 4 and 5
compare the conventional AM system and the integrated AM system results. Fig. 4(a) shows that the physical states of both AM
systems were very close and both met the threshold. However,
as highlighted previously, the conventional AM system failed to
meet the water network reliability in some specific areas. The integrated system successfully met the reliability threshold for all
areas. The integrated AM system resulted in a 5% lower replacement value [Fig. 4(b)], reflecting the close physical states of both
systems. Figs. 5(a and b) show the savings in terms of user costs
and lifecycle costs, respectively. The integrated AM system resulted in 50% savings in user costs compared with the conventional
AM system, implying both fewer interventions and less delay time
for service disruptions due to integrating the interventions of both
networks sharing the same spatial location. Moreover, the integrated AM system saved 33% in lifecycle costs compared with
the conventional AM system, which represents monetary savings
resulting from combining the intervention actions of both networks
due to the existence of common activities, i.e., traffic control
04017091-9
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Fig. 4. (a) Conventional versus integrated PS and RV comparison results—physical state (%); (b) conventional versus integrated PS and RV
comparison results—replacement value ($)
devices, excavation and backfilling of areas, site reinstatement, and
so on.
Revisiting Schedule Results
In addition to the overall KPI comparison, specific conclusions
were drawn through analyzing the section-based results of the integrated AM system versus the conventional AM system interventions; the integrated solution showed better and more efficient
planning in terms of the number of revisits carried out for the same
section (Fig. 6). The integrated interventions were favored over the
conventional-based interventions in most cases. The revisiting
schedule represents the frequency of carrying out interventions
© ASCE
in the same section throughout the planning horizon. The integrated
AM was more efficient in more than 90% of the sections, where the
integrated AM system carried out fewer intervention due to integrating the road and water networks, resulting in savings in terms
of interventions and user costs as well as less end-user service
disruption.
Conclusions and Future Work
The majority of the existing infrastructure AM frameworks in
the literature focus on the development of decision-making frameworks for conventional AM systems. However, because of spatial
04017091-10
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Fig. 5. (a) Conventional versus integrated UC and LCC comparison results—user costs ($); (b) conventional versus integrated UC and LCC
comparison results—lifecycle costs ($)
adjacencies and interdependencies among infrastructure systems,
some researchers have followed a more holistic view while
planning and scheduling intervention activities. This paper presents
an innovative integrated AM framework that can be used for interventions scheduling decision-making. It combines metaheuristics,
dynamic programming, and non-preemptive goal optimization
procedures to undertake the trade-off between bringing forward
or delaying infrastructure intervention activities. Furthermore, it allows decision makers to quantify the impact of their coordination
decisions and compare them with conventional AM systems. This
framework can be used to evaluate the coordination options in
urban areas that have multiple infrastructure systems within the
right-of-way. The use of metaheuristics and dynamic programming
© ASCE
significantly reduces the search space and allows the framework to
be scaled either to include more than two infrastructure systems or
to extend the planning horizon. The results of the implementation
case study show great savings by the integrated AM over the conventional AM in terms of user costs and LCC. Furthermore, it
proves that the integrated AM is 90% more efficient than the conventional AM in terms of section intervention frequency.
Despite the capabilities and flexibility of the system, future
work is underway to address some of the limitations through
extending the framework by (1) incorporating more than two infrastructure systems at the multiple-system level to maximize the coordination benefits, (2) including a GIS-based spatial model to take
the spatiotemporal perspective into consideration, (3) investigating
04017091-11
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
Fig. 6. Conventional versus integrated revisiting schedule
the use of other multiobjective optimization techniques, (4) enhancing the weight selection process through the analytical hierarchal
process to utilize expert opinions in the integration model or conducting sensitivity analysis to show the difference in outputs among
different weighting scenarios, and (5) accounting for other integration benefits as a part of the planning and decision-making process.
Data Availability Statement
All data generated or analyzed during the study are included in the
published paper.
References
Abu-Samra, S., Osman, H., and Hosny, O. (2017). “Optimal maintenance
and rehabilitation policies for performance-based road maintenance
contracts.” J. Perform. Constr. Facil., 10.1061/(ASCE)CF.1943-5509
.0000928, 04016075.
Amador, L., and Magnuson, S. (2011). “Adjacency modeling for
coordination of investments in infrastructure asset management.”
Transp. Res. Rec., 2246, 8–15.
Carey, B., and Lueke, J. (2013). “Optimized holistic municipal right-ofway capital improvement planning.” Can. J. Civil Eng., 40(12),
1244–1251.
City of Kelowna. (2016). “Kelowna map viewer.” 〈http://maps.kelowna.ca
/public/mapviewer/〉 (Jan. 18, 2017).
Daniels, P., Ellis, D., and Stockton, W. (1999). Techniques for manually
estimating road user costs associated with construction projects, Texas
Transportation Institute, College Station, TX.
De la Garza, J., Akyildiz, S., Bish, D., and Krueger, D. (2011). “Networklevel optimization of pavement maintenance renewal strategies.”
Adv. Eng. Inf., 25(4), 699–712.
DiSivo, M., and Ladiana, D. (2011). “Decision-support tools for municipal
infrastructure maintenance management.” World Conf. on Information
Technology, Elsevier, Amsterdam, Netherlands, 36–41.
Elbeltagi, E., and Tantawy, M. (2008). “Asset management: The ongoing
crisis.” Proc., 2nd Conf. on Project Management, Saudi Council of
Engineers, Riyadh, Saudi Arabia, 1–9.
Elsawah, H., Guerrero, M., and Moselhi, O. (2014). “Decision support
model for integrated intervention plans of municipal infrastructure.”
© ASCE
Proc., 2014 Int. Conf. on Sustainable Infrastructure Creating Infrastructure for a Sustainable World, ASCE, Reston, VA, 1039–1050.
Evolver version 7.0 [Computer software]. Palisade, Ithaca, NY.
FCM (Federations of Canadian Municipalities). (2016). “The Canadian
infrastructure report card, Ottawa, Canada.” 〈http://www.canada
infrastructure.ca/en/〉 (Jan. 18, 2017).
GAMS/CPLEX version 12 [Computer software]. GAMS Development
Corporation, Fairfax, VA.
Halfawy, M. (2008). “Integration of municipal infrastructure asset management processes: Challenges and solutions.” J. Comput. Civ. Eng.,
10.1061/(ASCE)0887-3801(2008)22:3(216), 216–229.
Halfawy, M. (2010). “Municipal information models and federated software architecture for implementing integrated infrastructure management environments.” Autom. Constr., 19(4), 433–446.
Halfawy, M., and Figueroa, R. (2006). “Developing enterprise GIS-based
data repositories for municipal infrastructure asset management.” Joint
Int. Conf. on Computing and Decision Making in Civil and Building
Engineering, Springer, New York, 1684–1693.
Hegazy, T., and Elhakeem, A. (2011). “Multiple optimization and segmentation technique (MOST) for large-scale bilevel life cycle optimization.”
Can. J. Civil Eng., 38(3), 263–271.
Houston, J. (2014). “Complex infrastructure projects made simple.” Conf.
on Pipelines: From Underground to the Forefront of Innovation and
Sustainability, ASCE, Portland, OR, 358–366.
InfraGuide. (2003a). An integrated approach to assessment and evaluation
of municipal roads, sewer and water networks, National Guide to
Sustainable Municipal Infrastructure, Ottawa.
InfraGuide. (2003b). Coordinating infrastructure works, National Guide to
Sustainable Municipal Infrastructure, Ottawa.
InfraGuide. (2003c). Planning and defining municipal infrastructure
needs, National Guide to Sustainable Municipal Infrastructure, Ottawa.
InfraGuide. (2004). Managing infrastructure assets, National Guide to
Sustainable Municipal Infrastructure, Ottawa.
InfraGuide. (2006). Managing risks, National Guide to Sustainable
Municipal Infrastructure, Ottawa.
ISO. (2014). “Asset management—Overview, principles and terminology.”
ISO 55000, Geneva.
Khan, Z., Moselhi, O., and Zayed, T. (2015). “Identifying rehabilitation
options for optimum improvement in municipal asset condition.”
J. Infrastruct. Syst., 10.1061/(ASCE)IS.1943-555X.0000220, 04014037.
Kimutai, E., Betrie, G., Brander, R., Sadiq, R., and Tesfamariam, S. (2015).
“Comparison of statistical models for predicting pipe failures:
Illustrative example with the city of Calgary water main failure.”
04017091-12
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Downloaded from ascelibrary.org by University Of Florida on 10/28/17. Copyright ASCE. For personal use only; all rights reserved.
J. Pipeline Syst. Eng. Pract., 10.1061/(ASCE)PS.1949-1204.0000196,
04015005.
Matthews, S. H. (2003). “Life cycle costing and reliability.” 〈http://
slideplayer.com/slide/5063156/〉 (Sep. 11, 2017).
Mirza, S. (2009). “Canadian infrastructure crisis still critical.” 〈https://www
.renx.ca/PDF2/09-Mizra-ppp.pdf〉 (Jan. 18, 2017).
NAMS (NZ Asset Management Support). (2015). International infrastructure management manual, 5th Ed., Wellington, New Zealand.
Oh, J.-S., Kim, H., and Park, D. (2011). “Bi-objective network optimization
for spatial and temporal coordination of multiple highway construction
projects.” KSCE J. Civil Eng., 15(8), 1449–1455.
Osman, H. (2015). “Coordination of urban infrastructure reconstruction
projects.” Struct. Infrastruct. Eng., 12(1), 108–121.
Ouyang, M. (2014). “Review on modeling and simulation of interdependent critical infrastructure systems.” Reliab. Eng. Syst. Saf., 121, 43–60.
Piñero, J. C., and de la Garza, J. M. (2003). “Issues related to the assessment
of performance-based road maintenance contracts.” 4th Joint Int. Symp.
on Information Technology in Civil Engineering, Towards a Vision for
Information Technology in Civil Engineering, ASCE, Nashville, TN.
Qin, X., and Cutler, C. (2013). “Review of road user costs and methods,
South Dakota Univ., Vermillion, SD.
Rashedi, R., and Hegazy, T. (2014). “Capital renewal optimisation for
large-scale infrastructure networks: Genetic algorithms versus advanced
mathematical tools.” Struct. Infrastruct. Eng., 11(3), 253–262.
Rinaldi, S., Peerenboom, J., and Kelly, T. (2001). “Identifying, understanding, and analyzing critical infrastructure interdependencies.” IEEE
Control Syst. Mag., 21(6), 11–25.
Schniederjans, M. (1995). Goal programming: Methodology and applications, Kluwer Academic, Boston.
© ASCE
Scott, S., Molenaar, K., Gransberg, D., and Smith, N. (2006). “Best-value
procurement methods for highway construction projects.” Project No.
10-61, National Cooperative Highway Research Program, Transportation Research Board, and National Research Council, Washington, DC.
Shahata, K., and Zayed, T. (2010). “Integrated decision-support framework
for municipal infrastructure asset.” Conf. on Pipelines: Climbing
New Peaks to Infrastructure Reliability: Renew, Rehab, and Reinvest,
ASCE, Keystone, CO, 1492–1502.
Shahata, K., and Zayed, T. (2016). “Integrated risk-assessment framework
for municipal infrastructure.” J. Constr. Eng. Manage., 10.1061
/(ASCE)CO.1943-7862.0001028, 04015052.
Statistics Canada. (2015). “Canadian demographics at a Glance: Population
growth in Canada.” 〈http://www.statcan.gc.ca/pub/91-003-x/2007001
/4129907-eng.htm〉 (Jan. 18, 2017).
Texas Department of Transportation. (2016). “Road user costs.” 〈http://
www.txdot.gov/inside-txdot/division/construction/road-user-costs.html〉
(Jan. 18, 2017).
Tong, T., Tam, C., and Chan, A. (2001). “Genetic algorithm optimization
in building portfolio management.” Constr. Manage. Econ., 19(6),
601–609.
Tscheikner-Gratl, F., Sitzenfrei, R., Rauch, W., and Kleidorfer, M. (2015).
“Integrated rehabilitation planning of urban infrastructure systems using a street section priority model.” Urban Water J., 13(1), 28–40.
Wood, N. (2015). “To fix Canada’s infrastructure, billions are just drops
in the bucket.” 〈http://www.cbc.ca/news/politics/canada-election-2015
-liberals-infrastructure-deficits-1.3206550〉 (Jan. 18, 2017).
Zimmerman, R. (2010). “Social implications of infrastructure network
interactions.” J. Urban Technol., 8(3), 97–119.
04017091-13
J. Constr. Eng. Manage., 2018, 144(1): 04017091
J. Constr. Eng. Manage.
Документ
Категория
Без категории
Просмотров
2
Размер файла
1 620 Кб
Теги
29co, 0001402, 28asce, 1943, 7862
1/--страниц
Пожаловаться на содержимое документа