Department of Civil and Environmental Engineering
University of Nevada, Reno (UNR)
Model-Based System for Rapid Post-Disaster Health Monitoring and Damage Detection of Civil Infrastructures
(U.S. Provisional Patent 62/040,932)
What does this system do?
This system provides an innovative, automated, on-site technology for the rapid structural health monitoring and damage identification of critical civil and lifeline structures (e.g., critical buildings, bridges, dams, industrial facilities, etc.) following a catastrophic event – specifically, earthquakes. The proposed system monitors and identifies the occurrence, location(s), and extent of damage throughout the structure using a physics-based model updating approach.
Why is this system important?
This technology provides a robust, accurate, and cost-efficient solution for rapid health monitoring and damage identification of critical civil infrastructure. Following a disastrous event, the system can provide complete information about the state of experienced damage in the structure of interest with high spatial resolution. This technology can be further used to assess the serviceability/functionality status and the remaining useful life of civil infrastructures, which is vital for emergency management and decision-making procedures to reduce fatality-risk during the critical period of time following a disastrous event, as well as to accelerating disaster recovery and rehabilitation.
How is this done today and what are the limits of current practice?
Current systems for structural health monitoring use a sensing network and a basic data analyzer to measure different response parameters of the structure. When any of the response parameters exceed a predefined threshold, the system automatically sends out warning signals (see Figure 1). This type of systems suffers from the following important drawbacks:
1. This type of systems only measures the global responses of the structure (e.g., peak interstory drifts), which reveal very little if any detailed information about damage sustained to the structure or the structural serviceability. This type of systems is susceptible to have false alarms, as a response parameter of interest may exceed the predefined threshold while the structure has experienced none or minor damages. Or, the structure may undergo severe level of damages or threats, while the monitored response parameters have not exceeded the thresholds.
2. Since this type of systems is not model-based, it cannot provide any detailed information on the location, type, severity, and extent of damage. As a consequence, this type of systems is only useful to determine whether the building should be evacuated for further inspections or can continue operation. The system, however, cannot have any impact on the inspection and therefore, costly and time-consuming visual inspections will be required as if the structural health monitoring system did not exist.
3. Increasing the accuracy of this type of systems requires deployment of a dense sensing network (i.e., using large numbers of different sensors to monitor the response of all parts, sub-assemblies, and components of the structure), which is an impractical solution. The dense sensing network results in high installation and maintenance costs and reduced system robustness due to possible sensor mal-functioning or erroneous reading.
2010 - present
2010 - present
Figure 1: Current system for structural health monitoring
The performance and robustness of the abovementioned type of systems can be improved by using linear system identification or modal identification methods to further process the data obtained from sensors at a deeper level. The modal parameters of an equivalent linear elastic viscously damped model of the structure are identified using low-amplitude (e.g., ambient) vibration data before and after a potentially damaging event. By comparing the modal parameters before and after the event, this approach can identify the occurrence of damage in the structure as a change in modal properties. This modified structural health monitoring system also suffers from some important deficiencies:
1. The changes in modal parameters of an equivalent linear elastic viscously damped model can be used to detect the occurrence of damage in the structure; however, it cannot be used to precisely determine the location, extent, and type of damage throughout the structural system. As a consequence, this system is also only useful to determine whether the building should be evacuated or can continue operation without any impact on the inspection process.
2. The system and modal identification methods approximate the structure as an equivalent linear elastic model. Linearity is an idealization of the behavior of real structures, which are intrinsically nonlinear from the onset of loading, even when subjected to low amplitude excitations. This limiting assumption reduces the accuracy of modal identification-based methods.
3. The low-amplitude (e.g., ambient) vibration data used for modal identification provide information about loss of effective stiffness; but, they are unable to reveal information about loss of strength (the most important manifestation of structural damage) and history of the response nonlinearities and related damages experienced by the structure. Therefore, this type of methods can sometimes incorrectly evaluate the state of damage in the structure.
Currently, the only applicable and trustworthy method for structural health monitoring and damage assessment is visual screening and inspection, which is a subjective, time-consuming, and extremely expensive procedure. An accurate inspection also sometimes requires destruction of architectural and nonstructural components, as well as removal of contents or equipment, and often cannot be performed during the brief, critical time following a catastrophic event.
What is novel in this new system as compared to existing technology?
Our system measures the input-output data during a damage-inducing event using a limited number of sensors (usually accelerometers) installed throughout the structure, and then uses the data to train and update an advanced mechanics-based nonlinear FE model of the structure through an innovative stochastic model-updating framework to reduce its associated uncertainties. Following a disastrous event, the system automatically interrogates the updated model for accurate identification (not only detection, but also localization, classification, and quantification) of damage in the critical civil structure and sends this information out through wired and/or wireless networks. The detailed information provided by this system can be further utilized to facilitate post-event retrofit plans.
Figure 2: Proposed model-based system for structural health monitoring and damage detection
(The red box shows the novelty of the proposed system compared to the existing art)
Inventors and developers: Hamed Ebrahimian, Rodrigo Astroza, Joel P. Conte.
1- H. Ebrahimian, R. Astroza, and J.P. Conte, "Extended Kalman Filter for Material Parameter Estimation in Nonlinear Structural Finite Element Models using Direct Differentiation Method," Earthquake Engineering and Structural Dynamics, 44(10), 2015, 1495-1522.
2- H. Ebrahimian, R. Astroza, and J.P. Conte, "Parametric Identification of Hysteretic Material Constitutive Laws in Nonlinear Finite Element Models using Extended Kalman Filter," Department of Structural Engineering, University of California San Diego, Structural Systems Research Report SSRP–14/06, La Jolla, CA, 2014.
3- R. Astroza, H. Ebrahimian, and J.P. Conte, "Material Parameter Identification in Distributed Plasticity FE Models of Frame-type Structures using Nonlinear Stochastic Filtering," ASCE Journal of Engineering Mechanics, 141(5), 2014, 04014149, DOI: 10.1061/(ASCE)EM.1943-7889.0000851.