Making Sense of Our Civil Infrastructure

The U.S. infrastructure is a trillion dollar investment, defined broadly to include road systems and bridges, water distribution and wastewater collection systems, water and wastewater treatment plants, power distribution systems, telecommunication network systems, commercial and industrial facilities, etc. Civil and Environmental Engineers play a major role in creation and management of our critical infrastructure. At Carnegie Mellon, our students learn that CEE researchers are at the forefront of a national movement to bring much greater use of sensing and information and communication technology to the construction and management of this infrastructure. We describe some of these research efforts in this article.

Figure 1: Laser scan of Phipps Conservatory

Providing Situation Awareness on Construction Projects

Buildings and other types of built infrastructure, such as bridges, are getting more and more complex in terms of the geometries and integrated subsystems used, and as such have become more difficult to construct properly. Yet most of the time the engineers managing these projects rely on infrequent and spatially sparse data being collected using mostly manual methods in the field, such as transit-based surveying, to assess the as-built and as-is conditions of the construction project and to make decisions accordingly. Possessing a limited understanding of the 3D as-built conditions during construction can potentially result in missing deviations and construction defects that actually exist in projects. Similarly, limited understanding of the 3D as-is conditions that exist during the lifetime of infrastructure, such as a highway bridge, can result in not being able to assess adequately the conditions of the infrastructure effectively. Professor Burcu Akinci and her colleagues and students are exploring a variety of sensing and tracking technologies, such as laser scanners and radio frequency identification (RFID) tags, that can help increase situation awareness. Laser scanners are able to capture millions of points on the surfaces of the infrastructure being constructed such that a very detailed survey of the three-dimensional geometry is captured (see Figures 1 & 2). RFID tags are used to keep track of the location and the actions taken on high-value engineered systems being stored and installed on a construction site. Akinci and her team of researchers have been focusing on assessing the capabilities of such technologies in capturing the data needed by engineers. They have also been developing approaches to process automatically the data collected so as to help engineers by providing proactive quality control during construction and providing proactive identification of deterioration and damage during the life-cycle of these infrastructure systems.

 

Figure 2: Laser scan of highway bridge

Determining Site-Specific Seismic Design Loads

The increasing complexity of buildings, bridges and other lifelines requires that careful attention also be given to the design process, especially under the action of transient loads, such as those due to wind, man-made changes, and earthquakes. Since a knowledge of the predicted ground motion to which a structure will be subjected during its lifetime is a key first step for determining the seismic design loads, and the seismic ground motion at a site depends strongly on the in-situ material properties of soil deposits and the deeper geological structure, it is essential to be able to characterize accurately the material profiles of spatially heterogeneous individual sites and complex basins alike. Professor Jacobo Bielak and his colleagues and students are integrating recently developed powerful inversion techniques with earthquake records and data from in-situ tests to perform high-fidelity site and basin characterization. They have already verified their inversion algorithms with pseudo-data from earthquake simulations on cross-sections of the Los Angeles basin derived from the SCEC Southern California Velocity model, as shown in Fig. 4, which depicts a prior model (left) used as an initial guess, along with the converged model (middle). The converged model is able to reproduce even fine features of the target model (right). To begin implementing this methodology in practice, they plan to conduct field tests this fall in a small valley in California, with support of a grant from the NEES program of NSF, using a large field shaker from the University of Texas, called Liquidator, to generate seismic motion that will be recorded by a large number of sensors (geophones) both on the free surface and inside boreholes. They will also be recording earthquake ground motion from small earthquakes with sensors (accelerometers) of the USArray component of Earthscope to augment the dataset from the field tests. This is a four-year project that, if successful, stands to revolutionize the way site characterization is performed, by removing many of the limitations inherent to current in-situ seismic inversion approaches, and by offering, for the first time, systematic tools for performing full three-dimensional site characterization.

Figure 4. Stages of a seismic inversion, starting from a prior model (left). The target model is on the right and the converged model in the middle; scale represents shear wave velocity in m/s

Monitoring our Drinking Water Distribution Systems

Drinking water distribution systems are another type of critical infrastructure, necessary for protection human health and economic development. However, they are monitored using simple methods including grab samples (i.e., samples taken from the system at various locations) and pressure and flow gauges. Real-time water quality monitoring for disinfectant residual (chlorine) or specific pathogens or chemical contaminants is in the development stage. Use of these sensors to manage and control the system as well as to detect intentional attacks requires careful attention to the placement of the sensors as well as an understanding of the variability and inherent uncertainties in the data they will return. Professor Jeanne VanBriesen and her colleagues and students, members of the Center for Water Quality in Urban Environmental Systems (Water QUEST), are focusing on optimized placement and data interpretation of sensors within drinking water distribution systems. They are exploring methods to use the data for real-time system control and more precise disinfectant dosing. Further, they are investigating methods to identify problems in the system that might result from operational errors, infrastructure failures, or intentional system attack. They envision an intelligent infrastructure that links sensors in the water distribution system with a data store and a model of the system to enable detection of anomalous system behavior and enhanced decision making to solve system problems.

 

Figure 3. Model Water Distribution System

Monitoring Our Sewer Systems

Another type of infrastructure being studied in the department is the wastewater collection system. The inspection and condition assessment of sewer pipelines has been impacted by two complementary but at present unintegrated developments. The first development, driven by EPA and state consent decrees, is an unprecedented increase in the use of closed-circuit television (CCTV) and the storage and mapping of this data in a geographical information system. This first development alone creates a rich dataset to be analyzed and an opportunity to improve decision support by using and developing spatial data analysis tools. The second development is alternative and complementary technologies for sewer inspection that overcome some of the limitations of the CCTV technology (see Figure 1 for one such example based on sonar-based sensing). These alternative technologies allow of deformation in the pipeline section and quantification of deposits. Professor Lucio Soibelman and his colleagues and students are attempting to create a framework that can harness these two developments, i.e. traditional inspection technology data and new technology development. This framework is intended to define the role, limitations, potential and means of integrating these inspections technologies, and provide guidelines for acquiring, storing, managing and analyzing these data. In addition to creating the framework described in the previous paragraph, Soibelman in collaboration with Professor Jim Garrett and their students are exploring, with support from RedZone Robotics, a means to automatically detect defects/anomalies and/or critical features using visual data acquired from sewer pipeline infrastructure inspection and condition assessment process.

Figure 1: Sonar inspection image (source: Redzone 2007, http://www.redzone.com/)

Monitoring our Oil and Gas Pipelines

With the ever expanding network of pipelines supporting the distribution of oil and natural gas in the US, there has been increasing demand for their continuous monitoring to support maintenance activities and for making intelligent decisions about these oil and natural gas distribution infrastructures. Professors Soibelman and Garrett with Adjunct Professor Hoon Sohn, are performing an in-depth theoretical and experimental study of the wave propagation in hollow cylinders actuated using innovative Macro Fiber Composite (MFC) transducers. This approach would then provide the basis to apply state-of-art damage detection techniques developed and tested successfully on plate like structures. The basic idea is that these patches act as both actuators and sensors and are able to pass the sent signals back and forth to each other so as to determine where local damage exists that disturbs the transmitted signal.

Developing New Sensors

Professor Irving Oppenheim, in collaboration with Professor David Greve from Electrical and Computer Engineering (ECE), has developed new sensors, including MEMS (microelectromechanical systems) devices, for civil infrastructure applications. Acoustic emission sensing is used to detect flaws or fatigue cracks in steel structures, and in a collaborative project with Professor Stephen Pessiki at Lehigh University the research team has developed four different MEMS devices to function as acoustic emission sensors. Their newest four-channel MEMS device, with its amplifiers, is housed in a 25x25x15 mm volume (see Figure 5) and is scheduled for testing on a railroad bridge in May or June 2007. In collaboration with Professor Hoon Sohn, and Professor Patrick Yue from ECE, an active transducer was developed for Lamb waves in steel plate structures, generating ultrasonic waves and then recording the echoes from boundaries and from flaws. The project was a major breakthrough, because the transducer mounted on the steel structure is totally passive (no wires, no batteries) and is powered by inductive coupling with a probe coil. This technology has been demonstrated on a steel plate girder in the CEE laboratories, and has been extended to full-size steel box girders. Oppenheim's research team has also developed an integrated circuit microsensor to measure chloride concentration. The sensor was first intended for use in monitoring chloride concentrations in concrete, but the technology is now envisioned to have considerable potential for sensing residual chlorine in drinking water systems, which is being studied in collaboration with Jeanne VanBriesen. The sensor developments in Oppenheim's research group have led to collaborations with many industry participants including Bombardier, Bosch, Physical Acoustics Corporation, TISEC, and WavesInSolids.

 

Figure 5: CMU four-channel MEMS system, 25x25x15 mm, 2007

The future of infrastructure systems will include significantly more information – generated during construction and operation and managed by intelligent control systems as well as human engineers and operators. Much more condition and usage data will be automatically and continuously collected, processed and converted into useful information about our infrastructure systems. This information will inform our decisions, from prioritization or repair and replacement to rapid response in the event of an intentional attack on these systems.

Several centers with significant CEE leadership and involvement are performing research in this area: WaterQUEST (previously mentioned) and CenSCIR, the Center for Sensed Critical Infrastructure Research, co-directed by Professor Jim Garrett and Professor Jose Moura from ECE.

Civil and environmental engineers are at the forefront of developing and evaluating the new technologies necessary to create this “sensed” infrastructure. At Carnegie Mellon, we are focused on identifying the roadblocks to implementation and on training the next generation of engineers to design and manage the intelligent infrastructure of the future.