Principal Investigator

Burcu Akinci

Status

Completed

Start Date

Jan. 1, 2018

End Date

Dec. 31, 2020

Project Type

Research Applied

Grant Program

FAST Act - Mobility National (2016 - 2022)

Grant Cycle

2017 Mobility21 UTC

Visibility

Public

Abstract

Civil infrastructure systems, such as roadways, pipelines and buildings, are among the largest investments in any
society. Most of these infrastructure systems in the U.S. were built during World War II era and they have
exceeded their design lifespans. Typically, infrastructure systems are located in close proximity, and are deployed
underneath the urban roadway networks. When they breakdown, there is a need to cut through roadway
pavements to access them - which in turn creates significant mobility implications to motorists. All of these
indicate a tightly-coupled nexus of interdependencies amongst underground utility/infrastructure systems,
roadways and motorist mobility. Most of the current approaches in managing these relationships and
corresponding disruptions are reactive; resulting in longer times of impact, higher costs and significant mobility
issues. There is an opportunity to rethink the current reactive approach and make it much more proactive and
effective. The proposed research targets that through foundational works towards the development of; (1)
predictive infrastructure maintenance operations that predict and coordinate major repairs amongst different
infrastructure systems to minimize disruptions to a community; and (2) dynamic incentive/disincentive models
provided to motorists to proactively manage and increase their mobility under a given maintenance related
disruption. The proposed approach includes ethnographic studies around existing planned disruptions through
maintenance, development of analytics for predictive maintenance and data-driven agent-based approach for
modeling motorist behaviors under different incentives/disincentives and disruption situations, and development
of an integrated analysis framework. If successful, the proposed research will provide the foundational work
towards a first of a kind integrated predictive coordinated maintenance and mobility management framework.
The research team will evaluate the framework through assessing how it reduces delays on the road network
segments and travel time for standard origin destination pairs, and how it enables time and money savings for
roadwork and utility companies through predictive and coordinated maintenance activities.    

Description

Motivating Vignette: Forbes Avenue between Morewood and South Craig Street at Pittsburgh, was shut down
unexpectedly on February 28th because of a 20 inch water main break. This required a new pipeline to be
installed and a major road construction. As a result, that portion of Forbes Avenue was closed for more than one
month and reopened on April 2nd. The closure impacted about 19,000 drivers, who uses that section of the road on
a daily basis, and the traffic got diverted onto much smaller streets, creating a major mobility challenge for all
motorists.
This vignette shows how tightly coupled our underground infrastructure systems, roadways and the mobility of
the motorists are. Unfortunately, this situation is not uncommon and is occurring at higher frequencies with the
aging infrastructure. A major problem with such situations is that in most of the cases that disruptions occur and
are managed in a reactive manner resulting in longer times of impact, higher costs and significant mobility issues.
There is an opportunity to rethink the current modus of operands and make it much more proactive and effective.
The proposed research targets that by doing foundational works towards the development of; (1) predictive
infrastructure maintenance operations that predict and coordinate major repairs amongst different infrastructure
systems to minimize disruptions to a community; and (2) dynamic incentive/disincentive models provided to
motorists to proactively manage and increase their mobility under a given maintenance related disruption.
Introduction and Overview of the Project
Civil infrastructure systems, such as roadways, pipelines and buildings, are among the largest investments in any
society. Most of the infrastructure systems in the U.S. were built during World War II (Chou et al. 2007). They
have met or are close to meeting their design lifespans, and are in need of rehabilitation to continue serving the
needs of the community. Typically, such utilities are located in close proximity, and are deployed along the urban
roadway networks.
In order to provide new services or maintain the deteriorating infrastructure networks, utility companies have to
cut pavements to access the utilities buried beneath (Jung 2012). Any disruptions due to utility breakdowns and
maintenance activities have direct and significant mobility implications. Hence, there is a need to minimize road
and utility maintenance through coordination of such activities. At the same time, incentives (or disincentives) can
be used as a way to alleviate mobility issues whenever there is a maintenance activity. All of these suggest the
existence of a tightly-coupled complex systems of systems between utility systems, transportation networks and
the citizens of a city.
Frequent cut and reinstatement of pavements have been shown to place physical stress on the pavement, leading
to reduction of lifespans by as much as 7 to 12 years (Jensen et al. 2005). This has also resulted in uneven
pavements, poor ride quality and increased accidents due to poor road safety (Khogali and Mohamed 1999). This
is significant, especially given that 32% of America’s major roads are in less than mediocre condition (ASCE 2014).
The cost of traveling on deficient pavement is estimated to be $67 billion a year, or $324 per motorist due to
repairs and operating costs (ASCE 2016). Deficient pavements are observed more commonly in urban areas, with
almost half of the interstate vehicle miles traveled are estimated to be over deficient pavements compared to only
15% on rural interstates (ASCE 2016).
Knowing the existence of such issues, FHWA has published a manual on controlling and reducing the frequency
of utility cuts (Wilde et al. 2002). This manual outlines various policies relating to Right-Of-Way, further
classifying them according to incentive-based, fee-based and requirement-based policies. While such policies are
very much needed, the effective usage of them still needs to be determined.
In this proposed research, our goal is to do foundational work towards the development of novel approaches that
optimize the coordination of utility/infrastructure and transportation maintenance and repair activities and that
enable experimenting with different incentive/disincentive strategies to minimize the mobility impact of scheduled
maintenance activities. We are proposing to do this in an integrated manner so as to be able to develop a
framework that not only minimizes the mobility impact of maintenance activities, but also proactively manages
them through incentives/disincentives that would be dynamically generated for motorists in a given context. Theproposed activities will pave the way to a unique framework that approaches mobility issues in an integrated endto-
end manner and proposes strategies to proactively manage them.
Background Research:
Typical Utility Management Systems include a database of utilities, with extensions to Geographic Information
Systems (GIS) for spatial visualizations and spatial data objects. A schema representing the structure of data
objects in the database typically include details on types of utility conflicts, work times, responsible parties, etc.
The typical goal of such UMS is to regulate and streamline the flow of utility information, and enable information
access to relevant parties.
Prior research in the area of coordination of maintenance of infrastructure systems can be categorized into two: (1)
Identification of information needed and the corresponding representation models to support coordination of
maintenance activities (e.g., Halfawy 2008 and 2010, Ganeshan et al. 2001); (2) Developing a decision-support
system for utilities contractors, providers and municipal authorities to strategically combine maintenance
activities to reduce mobility issues and corresponding time delays to the motorists (e.g., Chou et al. 2007; Arudi et
al. 2000).
At the same time, while there is a large amount of prior work on modeling travel behavior, there are not many
studies on behavioral responses to major network disruptions, such as the Forbes Avenue case described above
(Giuliano and Golob 1998). An agent-based approach has been demonstrated to be effective in modeling the
travel behaviors under a network disruption, such as snowy and icy conditions on roadway networks during a
snowstorm (Gonzalez 2015)
While these prior research studies provide a good foundational work for our proposed research, they have not
considered maintenance activities and possible dynamic incentives/disincentives that could be provided to the
motorists based on their anticipated behaviors to the disruptions in an integrated manner. Hence, they did not
approach maintenance, mobility issues, and motorist behavior as a nexus of interdependent systems that need to be
managed as such.
Research Approach
We will approach the proposed foundational work from two complementary and integrated perspectives: (1)
Modeling of maintenance needs of different infrastructure systems, and development of an optimization approach
to manage those needs in a predictive manner; and (2) Modeling how motorists behave under different
incentive/disincentives and developing a framework that dynamically generates incentives/disincentives to
motorists under given roadway construction activities due to maintenance and repair of infrastructure systems.
We would take advantage of the prior work in these discrete areas and identify new data sources to develop
predictive maintenance capability and more accurate cost models associated with disruptions. The absence of
metering or tolling systems is not an impediment. For example, it is possible to consider delays imposed on
individuals as “prices” faced by individuals on account of the disruption. There are also new data sources, such as
social media (e.g., twitter, Waze), that are being used to facilitate coordination in the face of disruptions. These
methods combined with simulation models can be used determine such “prices” – akin to what is described in
Gonzalez, 2015.
City of Pittsburgh provides a great test bed to develop and test several proposed research activities since they are
already working on ways to coordinate infrastructure maintenance activities to reduce mobility disruptions. We
will work closely with the local utility companies (such a People’s Gas, America’s water, Alcosan, Duquesne light)
together with the City of Pittsburgh to start modeling the maintenance needs and developing an optimization
approach towards managing those needs. The PIs already have close relationships with some of these companies
and the City of Pittsburgh to be able to implement that. At the same time, we will develop traffic models and a
simulation platform to perform different incentive scenarios and their implications. We will formalize both
maintenance optimization and incentive models in a combined manner so as to be able to have an integrated
framework for proactively managing both infrastructure maintenance and mobility needs.

Timeline

Quarter 1
- Ethnographic studies of infrastructure repair activities and corresponding mobility challenges.
People’s Gas is planning to repair/replace 100 miles of their gas pipelines in the City of Pittsburgh
and its metropolitan area throughout Summer and Fall 2016. Each segment of repair will be done at
a different location – which provide the research team a great opportunity to identify and observe the
mobility and traveler behavior at different contexts.
An outcome of this effort is to develop an ontology of disruptive events caused by breakdown,
scheduled repair or emergency repair activities. With respect to mobility, we are interested in the
nature of the change to mobility caused by the event. For example, the closure of the segment of
Forbes Avenue resulted in loss of access to that segment for all vehicles that wanted to use it as a
thoroughfare. Limited access to buildings on Forbes Avenue was available to vehicles for pick up and
drop off. Pedestrians and bicycles had access to the segment during the course of the repair.
Quarter 2
- Analytics for predictive maintenance based on the data that is provided for gas pipelines. A key
challenge here is predicting the likelihood of failure associated with various segments of a buried
network infrastructure taking into account the correlations induced both by the network, joints that
make up the physical structure of the pipes and the physics of fluid flow.
- Associated with each segment of interest of the buried infrastructure, we will develop a mapping
between buried infrastructure segments and above ground infrastructure or roadway segments. In
many cases, it will be a 1-1 mapping (e.g., like the mapping between the 20” water mains and the
Forbes Avenue segment), in others it may involve more than one road segment.
Quarter 3
- Analytics for predictive maintenance based on the data that is provided for gas pipelines (continued).
- Development and calibration of an agent-based model of mobility behavior under disruption. There
are existing simulation models of traffic flow and our proposal is to model and develop estimates of
delays under different vehicular and public transportation vehicular use assumptions in the presence
of disruption.
- The mobility model will be used to understand the impact – spatial and temporal scope -- of
disruptions (from simple to more significant as per the ontology we will develop). A goal is to
develop use cases of correlated disruptions. For example, the water mains break may cause a problem
on a segment of Forbes Avenue and a Gas Leak could cause a lane restriction in a segment of Fifth
Avenue. While the buried infrastructure disruption may be independent events from a causal
standpoint, they induce significant changes to mobility since the two disruptions affect a common
shared infrastructure.
- Our primary measure of impact will be induced delay on road network segments with a focus on
vehicular commuters. These delays will be used as “prices” faced by commuters in our modeling
framework. In the future we will consider pedestrian and bicyclists among the users of the road
infrastructure.
Quarter 4
- Integrated analytics, optimization and proactive mobility management framework. This work will
seek to integrate the predictive analytics associated with maintenance with a mobility management
under disruption.
- The mobility management model will enable policy interventions to be evaluated against the use
cases of buried infrastructure induced disruptions derived from our ethnographic case studies. The
objective function will be both delays on road segments in the presence of policy interventions as well
as travel time for standard origin destination pairs (e,g, typical commutes such as into downtown in
the morning).
- Since metering and tolling have not been deployed in Pittsburgh, our approach will be to simulate
these policy interventions – information and routing decision support tools to enable consumers to
make local travel choices in the presence of delays induced by the disruption with the objective of
balancing traffic load over the transportation network.
- Testing and validation of integrated framework in retrospective cases to be developed based on the
ethnographic case studies

Strategic Description / RD&T

Deployment Plan

We have a virtual deployment plan based on the real data and cases to be developed through
ethnographic studies

Expected Outcomes/Impacts

Many of the expected accomplishments and metrics are described in Section 8. In this section, we provide a highlevel
overview and summary.
- Development of the first version of an integrated predictive coordinated infrastructure maintenance
and mobility management framework.
- Detailed case studies of mobility implications of network disruptions due to infrastructure
maintenance activities.
- Metrics: (1) Reduced mobility issues via virtual agent-based simulations/demonstrations by tracking
metrics such as delay on road network segments and travel time for standard origin destination pairs.
(2) Time and money savings to roadwork and utility companies through predictive and coordinated
maintenance activities.

Expected Outputs

TRID

Individuals Involved

Email	Name	Affiliation	Role	Position
bakinci@cmu.edu	Akinci, Burcu	CEE	PI	Faculty - Tenured
garrett@cmu.edu	Garrett, James	CIT Deans Office	Other	Faculty - Tenured
rk2x@cmu.edu	Krishnan, Ramayya	Carnegie Mellon Heinz College	Other	Faculty - Tenured
seanqian@cmu.edu	Qian, Sean	Carnegie Mellon University	Co-PI	Faculty - Untenured, Tenure Track
zhuoran1@andrew.cmu.edu	Zhang, Zhuoran	Carnegie Mellon University	Other	Student - PhD

Budget

Amount of UTC Funds Awarded

$100000.00

Total Project Budget (from all funding sources)

$100000.00

Documents

Type	Name	Uploaded
Data Management Plan	198_-_DMP_Proactive_management_of_mobili.pdf	Dec. 17, 2018, 8:02 a.m.
Progress Report	198_Progress_Report_2019-09-30	Oct. 1, 2019, 7:08 a.m.
Progress Report	198_Progress_Report_2020-03-30	March 31, 2020, 7:36 a.m.
Final Report	Final_Report_-_198_mm8wyKT.pdf	April 14, 2021, 4:01 a.m.
Publication	Information Technology and US Infrastructure	April 19, 2021, 7:59 a.m.
Publication	Representation requirements for laser scan based probabilistic condition assessment of bridges	April 19, 2021, 8 a.m.
Publication	Automated damage detection and structural modelling with laser scanning	April 19, 2021, 8:01 a.m.

Match Sources

No match sources!

Partners

No partners!