Dynamic identification of structures

As part of the seismic improvement is very important validate the effectiveness of a intervention through the use of a FEM structural software that allows to verify the goodness of the planned intervention. What does dynamic identification consist of and what is its purpose?

Dynamic identification of structures

Validating the effectiveness of a seismic improvement intervention allows to verify the quality of the planned intervention.

The model of the building under investigation is created starting from the construction drawings and from the investigation campaigns conducted on the materials that make up the structural elements (pillars and beams) and non-structural elements (infill). 

Due to the limited information on the structure, the model thus created does not always represent a sufficiently faithful representation of the real dynamic behavior of the structure. In order to improve the dynamic behavior of the structural model, a methodology called "Dynamic identification". 

The purpose of dynamic identification is to find the modal parameters of the building, i.e. natural frequencies, modal masses, damping and modal forms. The knowledge of these parameters allows to describe the dynamic response of the building, considering the linear and time-invariant behavior. On the basis of the modal parameters identified in the field, it is possible to refine the finite element model in order to obtain a correspondence between the modal parameters of the FEM model and the experimental ones. The appropriately modified FEM model is said to be "calibrated" and allows for an extremely more accurate dynamic simulation that is relevant to the real behavior of the structure. 

Dynamic identification consists in measuring the movement of a structure subjected to external forcing (known or not) and, thanks to data analysis algorithms, extrapolating the parameters of interest. In order to carry out a correct identification, the structure must therefore be suitably sensorized. The accelerometric sensors are placed in significant points that allow to extract the modal forms of interest. The sensors must be able to perceive even minimal vibrations of the structure. For this reason accelerometers with very high sensitivity are generally used.


Figure 1: a) FEM model of the building; b) first way; c) second way; d) third way. Strategic positions of the sensors for the identification of the first three ways can be the vertices of the last floor

Currently two different methodologies are used to carry out dynamic identification: experimental modal analysis (EMA) e operational modal analysis (OMA). The difference between the two methodologies consists in the different excitation of the structure. 

In the EMA, the input forcing to the structure is known and controlled. It can be supplied through vibrodine or through a dynamometric hammer. With vibrodine, stationary (stepped-sine) or quasi-stationary (sine-sweep) excitations can be performed in the frequency range of interest (generally the range 0.5-15 Hz is sufficient to identify the modes that dominate the response of the building) . With the hammer, on the other hand, the forcing is impulsive, and very high frequencies are excited (even over 1000 Hz, depending on the hardness of the tip used). The input, by means of direct measurements on the vibrodine or with a dynamometric hammer, and the output, by means of the accelerometers, are sampled and analyzed in frequency, obtaining the Frequency Response Function (FRF) of the structure. The modal models are extracted by modifying the modal parameters until the best-fit between the numerical and the experimental FRF is obtained, by means of a least squares optimization. 

In the OMA, on the other hand, the forcing in input to the structure is unknown and not measured. It generally consists of environmental excitement (eg wind, traffic, footsteps, etc.). In the case of Operational Modal Analysis it is assumed that the input is a white noise, that is a random signal with a constant spectrum over a wide range of frequencies. Although this hypothesis is not exact, it is necessary to be able to use the mathematical tools underlying this method. Also in the case of OMA the modal parameters can be estimated by applying best fitting methods.  

The two methodologies have in common: 

  • Need to install sensors in strategic points 
  • Need to avoid local vibration phenomena (for example, localized work close to a single sensor can ruin an entire set of data) 

The table below collects the comparison between the two methodologies.

Ultimately, in all cases where the EMA is feasible (possibility of installing vibrodine, reduced or no number of people inside the building) it is preferable to OMA because it provides more precise and more immediate results. In all other cases, AOM can be applied. 

In any case, the dynamic identification process allows to obtain a sort of dynamic “identity card” of the structure on which it is carried out. This allows, therefore, to evaluate the state of health of the structure and allows the calibration process of the numerical models that are essential for the study of the behavior of the building subjected to seismic forcing.


Author: Fabio Bolognesi

Related Articles