MS18.7 - System Identification and Damage Detection

Spider webs are multifunctional tools. Besides capturing prey, spiders monitor vibration in the web to localize prey. Spiders have an evolutionary benefit of improving monitoring, because better information processing leads to more caught prey, leading to higher chances for survival. Based on this, combined with the costly energy requirements of signaling, biologists observe that the spider’s web operates as an Extended Cognition of the spider's Central Nervous System. Here an Extended Cognition means that the web filters signals to make decoding information easier. Spider webs contain characteristic design patterns, such as an eccentric location of the central hub. This paper investigates how this eccentricity influences the dynamic response of web-like structures, thus enabling Extended Cognition. A numerical model of spider web-inspired structures is used to attain the Structural Health Monitoring related insights and new sensing approaches found from the signaling capabilities of spider webs. Modeling consists of modal analyses in FEM on web-like structures where a mass is iteratively moved across all nodes. This yields the natural frequencies of the structure depending on the mass location. Analysis shows that the natural frequency forms patterns based on the mass location, and more importantly, it shows how these patterns become more intricate by including eccentricity. The resulting maximum change in natural frequency grows as the eccentricity in the design grows. This shows that eccentricity influences the dynamic response such that damage identification is enhanced.

A common task in experimental structural analysis is to acquire data for parameter estimation of some parametric model. For example, in experimental modal analysis, such a model is the modal model, and the parameters are modal parameters. The classical optimal sensors placement (OSP) formulation is to find the sensors’ configuration that minimizes the error in the estimated parameters using the least number of sensors. The presented paper examines this formulation for modal analysis: It compares the uncertainty in estimated modal parameters obtained with different sensor configurations in controlled environments. A simulated experiment is performed, where a modal test of a simple mechanical system is conducted for optimal and “naïve” sensors’ configurations, and the measurement uncertainties are propagated through the modal parameters estimation algorithm to obtain the uncertainties in modal parameters. The results demonstrate the applicability of the OSP.

Leakage in buried pipes is the main source of wastage in water distribution systems. The issue of promptly pinpointing and fixing the leaks is thus of great interest to minimize water wastage. One way to pinpoint a leakage is to measure two acoustic signals, on the pipe, one each side of the leak and estimate its position from the delay of arrival between the signals, given the velocity of the fluid-borne wave that propagates from the leak along the pipe. This technique is limited by the necessity of access points to the buried pipe, by the distance from the leak and measurement points and a good estimate of the velocity of the fluid-borne wave. Moreover, it is necessary to have some a priori knowledge of the probable location of the leak before taking the measurements. To complement this approach, we propose an alternative technique based on surface vibration measurements in an area above the leak. The technique involves a “seismic camera”, which measures the acoustic signal generated by the leak on the ground surface with an array of geophones and analyses the data using steering vector array algorithms. Although the proposed concept may appear to be similar to existing acoustic cameras, it differs drastically. Acoustic cameras are developed based on the acoustic wave propagation in fluids in which only longitudinal pressure waves propagate at the local sound velocity. In soil two or more different wave-types may propagate with different phase velocities. Also, the pressure oscillation caused by waves in a fluid is an omnidirectional scalar quantity, whereas the velocity oscillation of particles in a solid is vectorial. In this paper, the concept of the seismic camera is described, then the effects of multiple wave types and the measurement direction are investigated numerically. A simulation of a monopole source inside an infinite thin cylindrical shell in an unbounded media is performed to assess the array signal filtering performance. Results from the numerical simulation show that a specific wave type (shear or longitudinal) dominates the vibration response depending on the frequencies radiated by the leak. These observations are also investigated experimentally with measurement of vibration radiated by a buried pipe excited by an open valve to simulate the leak. The results show that the use of array signal processing in vibration measurements around the leakage may be a useful addition to existing techniques in the location of leaks in buried water pipes.