Introduction

Structural Health Monitoring has become a mature field of research for the last ten years [1]-[7]. The reason of fast development of the field is an urgent of need of infrastructure operators to maintain the aging structures in the state assuring both the serviceability and safety.

SHM has its roots in the classical non-destructive testing methods, especially vibroacoustics and ultrasonics. Similarly to NDT methods, SHM needs to know the history of structural behaviour obtained from in-situ measurements within some time in order to assess the state of the structure. Unlike the NDT methods though, SHM aims at regular and automated monitoring of structures, which is supposed to replace scheduled human inspections on site in the near future. Thus the problems considered by researchers dealing with SHM include algorithms of damage identification on the one hand (software) and methods of reliable data acquisition and transfer on the other (hardware).

SHM can be divided into two major streams according to the methods of origin, namely low-frequency (non-ultrasonic) SHM, stemming from vibroacoustics and relying on vibration analysis and high-frequency (ultrasonic) SHM, stemming from ultrasonic diagnostics and relying on elastic wave propagation. The LF SHM has its applications mainly in civil and mechanical engineering while the HF SHM in aerospace engineering.

In terms of damage identification algorithms developed for SHM, several stages of damage recognition can be distinguished:
  • 1. detection (is there damage?)
  • 2. localization (where is the damage?)
  • 3. identification (how severe is the damage, what type of?)
  • 4. prognosis (how much useful lifetime remains?)
  • 5. self-healing (can the structure self-initiate repair procedures on site?)
Most frequently used sensors in SHM include acceleration and strain sensors, first of all well-known accelerometers and strain gauges but also optical fibres and piezoelectric patches recently growing in applications.

Vibration-based SHM

The LF SHM usually takes advantage of modal analysis [8]. Both the theoretical and experimental modal analyses are performed in order to identify proper structural parameters in the process of system identification [9] and subsequent model updating [10]. Once the numerical model has been tailored to experimental measurements, the LF SHM focuses on examining discrepancies between the reference and actual responses (signatures) of the structure. This can be done by solving an inverse problem [11] in the process of optimization (either sensitivity-based or using soft-computing methods e.g. genetic algorithms). The objective function usually involves eigenfrequency and eigenmode components. Damage is identified by the parameter change assigned to specific elements of the structure. An alternative to optimization is a pattern recognition approach which compares the current case with the already existing cases of a representative and numerous data base of damage cases [12].

Degradation of stiffness is commonly used as damage indicator. Except for eigenfrequencies and eigenmodes, other damage-sensitive features are used e.g. curvature [13] or energy-based criteria [14]. The most often investigated structures in LF SHM are bridges in civil engineering and rotating machinery in mechanical engineering.

Acceleration vs. strain measurements

When carrying out modal analysis, one frequently uses accelerometers for collecting structural responses [15]. As the raw acceleration signals are cumbersome to analyze directly, one transfers the data from the time to frequency domain by the Laplace or Fourier transform. Then the major modal parameters are identified in the frequency domain using e.g. the frequency response functions. Depending upon approach some data may be transferred back to the time domain by using inverse Laplace or Fourier transforms in order to enhance the frequency-based methods. Anyway, the basic response of the structure that is analyzed is acceleration. The commonly used sensors are piezoelectric accelerometers, which are easy-to-use and accurate devices.

An alternative to the acceleration measurements is to look at strains (deformation). An inherent advantage of this approach is that one measures a physical quantity directly. Another asset is that strains do not evolve in time as rapidly as accelerations, which makes the responses smoother and suitable for analysis in the time domain right away. A drawback of the strain measurements is certainly a more troublesome procedure of collecting data. The strain sensors must be well-attached to the surface of the structure in order to produce a meaningful reading. Therefore the installation of standard strain gauges [16] is demanding and time-consuming. Recently developed technologies of fibre optic [17] and piezoelectric patch sensors make the issue easier, although proper attachment of these sensors to the structure is essential for proper measurements.

References

  • [1] International Conference on Damage Assessment of Structures DAMAS: Pescara, Italy, 1995; Sheffield, UK, 1997; Dublin, Ireland, 1999; Cardiff, UK, 2001; Southampton, UK, 2003; Gdansk, Poland, 2005; Torino, Italy, 2007, Beijing, China, 2009
  • [2] North American Workshop on SHM: Stanford University, USA, every 2 years, 7th event in 2009
  • [3] European Workshop on SHM: Cachan, France, 2002; Munich, Germany, 2004; Granada, Spain, 2006; Kraków, Poland, 2008
  • [4] Structural Health Monitoring and Intelligent Infrastructure: Tokyo, Japan, 2003; Shenzhen, China, 2005; Vancouver, Canada, 2007; Zurich, Switzerland, 2009
  • [5] EVACES, Experimental Vibration Analysis for Civil Engineering Structures: Bordeaux, France, 2005; Porto, Portugal, 2007, Wrocław, Poland, 2009
  • [6] Structural Health Monitoring and Intelligent Infrastructure: Tokyo, Japan, 2003; Shenzhen, China, 2005; Vancouver, Canada, 2007; Zurich, Switzerland, 2009
  • [7] Noise and Vibration Engineering: Leuven, Belgium, every 2 years, 23rd event in 2008
  • [8] Ewins D. J. (2001) Modal Testing: Theory, Practice and Application, Taylor & Francis Group, 2nd edition
  • [9] Juang J-N. (1994) Applied System Identification, Prentice Hall PTR, Englewood Cliffs, NJ, USA
  • [10] Friswell M. I., Mottershead J. E. (1995) Finite Element Model Updating in Structural Dynamics, Kluwer Academic Publishers
  • [11] Friswell M. I., Mottershead J. E. (2001) Inverse Methods in Structural Health Monitoring, Proc. of the International Conference on Damage Assessmentof Structures DAMAS01, 25-28 June, Cardiff, UK, pp. 201-210
  • [12] Mujica L. E., Vehi J., Rodellar J., Kolakowski P. (2005) A Hybrid Approach of Knowledge-Based Reasoning for Structural Assessment, Smart Materials and Structures, vol. 14, issue 6, pp. 1554-1562
  • [13]Maeck J., De Roeck G. (1999) Dynamic Bending and Torsion Stiffness Derivation from Modal Curvatures and Torsion Rates, Journal of Sound and Vibration, vol. 225(1), pp. 153-170
  • [14] Fritzen C.-P., Bohle K. (2004) Damage Identification using a Modal Kinetic Energy Criterion and Output-Only Modal Data - Application to the Z24-Brigde, Proc. of the 2nd European Workshop on Structural Health Monitoring, 7-9 July, Munich, Germany, pp. 185-194
  • [15] Gautschi G. (2002) Piezoelectic Sensorics, Springer
  • [16] Window A. L. (Ed.) (1992) Strain gauge technology, 2nd Edition, Elsevier Applied Science
  • [17] Udd E. (Ed.) (1995) Fibre Optic Smart Structures, Wiley, New York