Download PDF Spring Bridge on Sustainable Smart Cities March 15, 2023 The world’s cities face increasing threats from natural disasters, aging infrastructure, traffic, and resource constraints. The articles in this issue examine smart infrastructure, sustainability, net zero carbon options, and autonomous driving, among other approaches to smart and sustainable cities. Smart Infrastructure for Smart Cities Friday, March 17, 2023 Author: Kenichi Soga How can the built environment be rehabilitated or created so that future generations benefit from smart infrastructure? Much of the nation’s infrastructure is aging and in poor condition, affecting safety, the economy, and quality of life. A variety of emerging technologies can enhance infrastructure to improve safety, resilience, sustainability, and equity. Challenges to Current Infrastructure Systems Reactive, damage-based management is ineffective. It takes a long time to build infrastructure, with construction timescales alone stretching from 2 to 10 years. As shown by the first row in figure 1, many infrastructure assets are designed for a service life of 100 years, even with deterioration due to material degradation, extreme temperature, and external loads. But deterioration can accelerate because of poor design or workmanship, construction problems, unforeseen stressors, and inadequate maintenance and repair—it’s worth noting that effects of changes in traffic mode, demand, or weather events are not currently considered in maintenance. Continuous retrofit, renovation, and adaptation are required during an infrastructure’s lifetime, and the high cost involved in upgrading and replacing leads to a desire to extend overall life, as illustrated by the second row in figure 1. The American Society of Civil Engineers (ASCE 2021) has estimated that the cumulative needs for US infrastructure—in the form of inspection, maintenance, repair, and replacement expenditures—could reach $5.9 trillion by 2029, but that the estimated available funding is only $2.59 trillion. There is thus a compelling case for extending the useful life of infrastructure. Assets also face external conditions that deviate from what was known or assumed by the planner or designer, such as population growth/decline, more frequent natural hazards due to climate change, fluctuating energy prices, and shifts in transport modes. These changes often occur several times during the life of infrastructure (the third row in figure 1). Ideally, infrastructure should be designed to both meet immediate needs and be adaptable to future demands throughout its lifetime. Past design philosophy, however, was based on current demand prediction, creating a substantial risk that the infrastructure will be inadequate or obsolete before the end of its expected period of operation. In addition, the covid-19 pandemic changed infrastructure demands as teleworking continues to transform residential and travel patterns. Adopting new mobility platforms and increasing automation and electrification will affect future infrastructure. Adaptation is no longer a choice but a requirement for sustainable living. Infrastructure must adapt to changes and threats that are here now. The need to improve the capability to predict, design for, and manage the life expectancy of infrastructure calls for smart infrastructure engineering with the sustainability, resilience, and equity of communities at its center. How can the built environment be rehabilitated or created so that future generations benefit from smart infrastructure? Stakeholders and Layers in Smart Infrastructure Engineers manage infrastructure safely and economically by dealing with the uncertainties of such assets’ life expectancy and performance during hazards. Infrastructure owners, on the other hand, are faced not only with loss of service and lifecycle costs but also uncertainties such as changes in demand, climate, policy, and environment (figure 2). And they must be responsive to stakeholders’ expectations. Engineers provide for infrastructure performance in the future based on understanding and predicting actual performance through sensing and modeling. Infrastructure owners need to know their asset’s projected service performance based on information about its anticipated behavior given by engineers. Both need to make decisions for short- and long-term performance based on information obtained from monitoring of the infrastructure. Their judgment is guided by solid evidence, and decisions are made assuming multiple alternatives. The key to realizing smart infrastructure is to verify that the link between asset behavior and service performance is well established. Figure 2 (right side) illustrates how smart infrastructure can be developed at different scales and layers. Engineers deal with the bottom three layers, and infrastructure owners deal with the top two: both are concerned with the asset layer. The layers—sensors and data collection, data analysis and interpretation, assets, and infrastructure system—are discussed in the following sections, with factors and questions to be considered for each. Sensors and Data Collection If sensor technology can be used for an extended period (equal to the infrastructure lifetime) with appropriate maintenance and replacements (as shown by the fourth row in figure 1), it will be possible to introduce the lifecycle approach in the civil engineering industry. The information required includes data quality and its degradation over time, survival rates of hardware and software components, and the associated error propagation and cost of component management. Some typical metrics to consider for whole-life sensing are the level that can measure the performance of infrastructure to make engineering decisions, robustness and reliability of sensors, frequency of data collection, and replaceability as newer sensors become available. Data Analysis and Interpretation Fundamental challenges in this layer lie in creating new models or modifying existing models that drive economic sensing requirements and sensor deployment for specific applications. For example, what resolution and sampling frequency are needed to detect long-term degradation effects, natural hazards, and harmful movements due to adjacent construction or climate change? Answers to this question will require new models that anticipate future stresses on the infrastructure and revision of existing models to correspond to what new sensors can measure. Following are typical questions to answer at this level: What data are needed to do this sensor-integrated modeling, and how should it be interpreted? Where should sensors be located, and at what time and spatial resolution? How can real-time information about physical assets inform usage strategies and future design? How can assets and sensor-integrated models be future-proofed against changing requirements and shocks? Furthermore, the quality of data from sensors and monitoring systems changes with time, and potential error propagation due to aging must be quantified. These questions will be answered by new models that focus on (i) data quality and its degradation over time, (ii) survival rates of hardware and software components and the associated error propagation, and (iii) costs of management and maintenance. Any gaps among the formats need to be identified, and good data transfer links are essential. Data linking may produce errors, which need to be quantified for accurate modeling and assessment of infrastructure performance. Assets ISO (2014) 55000 standards on asset management highlight the importance of the lifetime management of physical assets and of realizing value rather than minimizing cost. Asset value needs to be determined from a multistakeholder perspective. Asset owners face a multiperspective challenge that includes balancing cost and risk with decreasing funding and increasing regulation. Building information modeling (BIM) and digital twin techniques that manage design and construction information in a transparent manner can aid in tackling these challenges. Following are some typical questions to answer at this level: How must assets be operated, managed, and maintained to deliver their best whole-life value? What decisions are needed to support such operation, management, and maintenance? What information is needed to make those decisions? What new engineering design, construction, and maintenance processes need to be developed for an integrated adaptive infrastructure system (i.e., system of systems)? What kind of institutional objective utility/optimization is needed for asset owners? Cities and Infrastructure Systems Changes in physical infrastructure, transportation, utilities, and communications entail some adaptation for citizens. With the use of sensor data to make infrastructure more adaptable and resilient, this layer can reduce or eliminate inefficiencies in the provision of services while maintaining the integrity of city infrastructure systems. The ideal is that human behavior and infrastructure evolve together to enhance quality of life while supporting vibrant business, trouble-free transportation, and efficient, sustainable use of resources. Following are typical questions for this layer: How does infrastructure best serve its communities? How will infrastructure change the current spatial-temporal pattern of cities’ transportation networks, energy consumption, commercial activities, lifestyles, and environmental quality? What will be the cascading effects after a natural hazard? What kinds of policies and planning procedures best incentivize change in infrastructure design, construction, and use? Emerging Technologies for Smart Infrastructure Emerging technologies empower decision makers. And many new technologies related to materials, sensing, communication, and computing (Soga and Schooling 2016) may be used for smart infrastructure applications (table 1). To illustrate the maturity of myriad new technologies relevant to smart infrastructure, their current lifecycles are assessed and mapped using the Gartner (2022) Hype Cycle (figure 3). The Gartner chart illustrates five phases of the technology adoption process: the innovation trigger, peak of inflated expectations, trough of disillusionment, slope of enlightenment, and plateau of productivity. The vertical axis is a psychological measure of market expectation. Sensor Systems: Multiple Scales, Autonomy Integrating structural sensing, environmental sensing, and infrastructure usage can yield significant benefits. For example, the degradation of infrastructure is typically governed by cyclic thermal loading (expansion/contraction), changes in moisture conditions (e.g., humidity, flooding, groundwater pressures), and/or changes in use (e.g., heavier traffic, change in flow volumes and pressures). Integration and communication between long-term value (e.g., structural health, future hazards, degradation) and short-term value (e.g., operation, energy) may provide efficiencies and profoundly shift how infrastructure projects are managed and maintained. But there is a mismatch between the lifespan of infrastructure and that of sensor and digitized data management systems, which makes the concept of lifecycle-based asset management difficult to realize. Some currently used data may be from older sensors, some sensors may be embedded now but the data will be used in 10, 20, or 50 years. Intelligent sensor and data management systems must be designed for long lifespans or adaptable for replacement. Smart infrastructure requires a multiscale approach for sensing/monitoring coupled with modeling that uses the data collected. Figure 4 shows the capabilities of satellites, unmanned aerial vehicles (UAVs), and wireless sensors. The point-based wireless sensor networks are hampered by their limited numbers. The risk is that one does not have a sensor where a critical failure occurs. Aerial technologies such as satellite equipment (e.g., optical, InSAR), UAV-based lidar, and photogrammetry are producing point cloud options to measure changes in surface conditions over time, including deformations. The data from these systems can be used to evaluate the performance of an infrastructure system. But these technologies are not yet developed for the simplified user stage and are not commonly integrated with point-based sensors, limiting the value of both. New larger construction projects often have effective monitoring systems to help manage risks associated with construction, but most of these systems are generally removed or abandoned when construction ends. With guidance for better integration and coordination, some of these large investments could be sustained in lifelong monitoring systems. Automated construction technologies in the framework of a common data environment (CDE) can generate greater amounts of improved (more standardized) data, more economically and without causing extra burden to the limited human resources available. Autonomy during the operational stage may provide insights into how the built environment is functioning and empower new business models that leverage data to achieve unprecedented efficiencies in infrastructure systems. For example, a traffic analysis service uses in-vehicle navigation apps in smartphones. Digital Twins: From the Technical to the Sociotechnical A digital twin (DT) is a digital representation of a physical (infra)structure (Grieves and Vickers 2017). Using simulation, it enables data to be managed and analyzed to, among other things, understand how people interact with infrastructure systems. Technical Advantages Digital twins can increase the number of dimensions in BIM beyond 4D (3D + time) with additional performance indicators such as cost, lifecycle and maintenance information, sustainability, and safety (Boje et al. 2020; Ding et al. 2014). They also provide the opportunity for a better understanding of the infrastructure, subsurface, and sensor data through visualization using virtual and augmented reality. And they not only account for both the past performance and present state of the modeled infrastructure, but also are a basis for scenarios to predict likely performance, ensure preparedness, and enhance resiliency. To advance the ability to simulate natural hazard impacts on structures, lifelines, and communities, the SimCenter,[1] funded by the National Science Foundation and hosted by the University of California, Berkeley, provides next-generation computational modeling and simulation software tools, user support, and educational materials to the natural hazards engineering research community, using a new cloud-enabled open-source framework (Deierlein et al. 2020). Engineering innovation to develop resilient infrastructure was also the focus of the 2019 summer issue of The Bridge (O’Rourke 2019). Sociotechnical Applications Inequities exist in transportation, housing, urban heat island effects, flooding, energy, and water supply. The transition from existing aging infrastructure to zero-carbon infrastructure must be equitable in the face of growing environmental challenges. These problems cannot be solved without deeply considering the complex socioeconomic and political considerations that impact different communities at different scales. The definition of infrastructure is expanding, and should include organizational infrastructure (human interactions) and informal infrastructure (unplanned) in the physical and digital infrastructure framework. The complexity of the social decision processes involved in mobilizing change requires the creative use of DT technologies, such as a sociotechnical digital twin, which integrates models of physical infrastructure systems and virtual networks, including organization, community, and communication networks. Novel structural designs are required to integrate new materials that can eliminate or capture direct greenhouse gas emissions. Infrastructure systems need to be examined collectively as an integrated system using theoretical concepts such as system dynamics (Forrester 1969), complex adaptive systems (Holland 1992), and systems of systems (Maier 1998). Collaboration with humanities and social sciences experts is a must. Artificial Intelligence and Machine Learning Because the data collected by myriad sensing technologies are extensive, big data approaches are needed to leverage their strength. Machine learning (ML) and artificial intelligence (AI) combined with high--performance computing provide promising techniques to detect trends in high-dimensional data, which was not possible with traditional statistical techniques. This is particularly true for large-scale infrastructure with numerous data channels incorporating multiple measurement parameters, image-based sensing, or other noncontact sensing that generates large datasets. ML/AI can process infrastructure images collected from a drone to find patterns (classification) and anomalies in the surrounding area with relatively high precision. The image dataset can then be combined with other sensing data (e.g., from the structures) to provide a multi-perspective sensing dataset. Supervised learning is a powerful interpolation tool that can find complex patterns in high-dimensional data without predefined physical laws and assumptions. It may perform poorly, though, in extrapolation problems where the conditions are outside the training boundaries. This constraint may be solved by getting more data and continually expanding the training boundaries. However, this approach is applicable only if there are no catastrophic consequences due to prediction errors—such errors can lead to serious failure and unreliable predictions. Some models are hampered by overfitting and may perform reliably only within given training boundaries. A model that produces substantial errors due to a lack of generalization (i.e., inability to adapt to new data) or data perturbation (e.g., outliers, noises) cannot be tolerated. This is an important limitation of ML/AI for smart infrastructure applications. The focus needs to shift from prediction accuracy to prediction reliability by including probabilistic concepts and statistical tools such as bootstrapping and cross-validation. Materials for Net Zero Carbon Infrastructure The concept of smart infrastructure should not only include innovations in whole-life sensing and data analytics but also adopt innovations in materials and construction/maintenance processes. Future infrastructure systems must be designed to generate their energy or rely exclusively on renewable energy, realizing a net zero or negative carbon system. Innovations in self-healing and self-sensing materials have great potential to both extend the life of infrastructure and enhance the resilience of new infrastructure. Novel structural designs are required to integrate the distinct properties of new materials that can eliminate or capture direct greenhouse gas emissions. Methods of modular structural design and construction are needed to enable an adaptable infrastructure that can change with user and technology demands. With synergistic advances in these areas, future infrastructure systems will not only satisfy immediate needs but also be adaptable to evolving demands throughout their lifetime as part of a circular and net zero carbon economy. Conclusions The world is becoming more resource-poor, more connected, and more interdependent. The parameters that affect prosperity are also constantly evolving and spatially variable, contributing to uncertainty about the future. Infrastructure that is adaptable by design must involve the input of communities to enhance understanding of disparities and development of long-term solutions. Design philosophy has typically focused on current demand prediction, creating a substantial risk that the resulting infrastructure becomes inadequate or obsolete in a few decades (or less). Smart infrastructure can predict, design for, and manage its life expectancy by using emerging technologies such as digital twins, net zero or negative carbon materials, sensors, robotics, and new processes. Scientific and technological advances make it possible to generate and analyze data about how infrastructure and the environment are used by communities and to develop new ways that are equitable, sustainable, and resilient. Effective engineering service involves listening and responding to the interests and concerns of the people engineers serve. Smart infrastructure in smart cities should embrace this vision and pursue ways to realize it. Acknowledgments The author thanks the following individuals for helping develop the ideas described in this paper: Robert Mair (NAE) and Jennifer Schooling, Cambridge Centre for Smart Infrastructure and Construction; Matthew DeJong and Dimitrios Zekkos, Berkeley Center for Smart Infrastructure; Thomas O’Rourke (NAE), Cornell University; Louise Comfort, Center for Information Technology Research in the Interest of Society (CITRIS), UC Berkeley; Tracy Becker, Peter Hubbard, and Dayu Apoji, UC Berkeley; and Pingbo Tang (Carnegie Mellon University), ZhiQiang Chen (University of Missouri–Kansas City), and Mahmoud Reda Taha (University of New Mexico), ASCE Infrastructure Resilience Division’s Emerging Technologies Committee; and Cameron Fletcher, National Academy of Engineering. References ASCE [American Society of Civil Engineers]. 2021. 2021 Report Card for America’s Infrastructure. Reston VA. Boje C, Guerriero A, Kubicki S, Rezgui Y. 2020. Towards a semantic Construction Digital Twin: Directions for future research. Automation in Construction 114:103179. Deierlein GG, McKenna F, Zsarnóczay A, Kijewski--Correa T, Kareem A, Elhaddad W, Lowes L, Schoettler MJ, Govindjee S. 2020. A cloud-enabled application framework for simulating regional-scale impacts of natural hazards on the built environment. Frontiers in Built Environment 6:558706. Ding L, Zhou Y, Akinci B. 2014. Building information modeling (BIM) application framework: The process of expanding from 3D to computable nD. Automation in Construction 46:82–93. Forrester JW. 1969. Urban Dynamics. Pegasus Communications. Gartner. 2022. Gartner Hype Cycle. Online at https://www.gartner.com/en/research/methodologies/gartner- hype-cycle. Grieves M, Vickers J. 2017. Digital twin: Mitigating unpredictable, undesirable emergent behavior in complex systems. In: Transdisciplinary Perspectives on Complex Systems: New Findings and Approaches, eds. Kahlen F-J, Flumerfelt S, Alves A. Springer. Holland JH. 1992. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence. MIT Press. ISO [International Organization for Standardization]. 2014. ISO 55000:2014: Asset management — Overview, principles and terminology. Available at https://www.iso.org/standard/55088.html. Maier MW. 1998. Architecting principles for systems of systems. Systems Engineering 1(4):267–84. O’Rourke TD, ed. 2019. Engineering for Disaster Resilience. The Bridge 49(2). Soga K, Schooling J. 2016. Infrastructure sensing. Interface Focus 6(4):20160023. [1] https://simcenter.designsafe-ci.org/ About the Author:Kenichi Soga (NAE) is the Donald H. McLaughlin Professor and director, Berkeley Center for Smart Infrastructure, Department of Civil and Environmental Engineering, University of California, Berkeley.