Ifsttar PhD subject

French version
Detailed form :

Title : Detection of defects in civil engineering structures by vibratory response analysis using wavalet transform

Main host Laboratory - Referent Advisor   -

Director of the main host Laboratory   -

PhD Speciality Mécanique

Axis of the performance contract 2 - COP2017 - More efficient and resilient infrastructure

Main location Marne-la-Vallée

Doctoral affiliation UNIVERSITE PARIS-EST

PhD school SCIENCES, INGENIERIE ET ENVIRONNEMENT (SIE)

Planned PhD supervisor ARGOUL Pierre  -  Université Gustave Eiffel  -  LVMT

Planned financing Contrat doctoral  - Ifsttar

Abstract

To meet the increased need for monitoring aging structures and infrastructures, the study of their vibratory behavior in operational conditions, as part of Structural Health Monitoring (SHM), offers a cheap and promising means of damage detection. For this purpose, the Continuous Wavelet Transform (CWT) is a suitable tool, both in the study of the evolution of modal parameters, and by its ability to detect and characterize non-linear and non-stationary dynamic behaviors. The aim of this thesis is to contribute to the improvement of current analysis techniques based on this transform, and to the development of new methods.

In the case of transient or harmonic responses, existing solutions of modal analysis using wavelets are poorly adapted to multi-channel signals. A joint transform for these signals is therefore proposed, allowing a reduction of the relative noise and the definition of a single ridge, along with a criterion of normalization of the instantaneous mode shapes in the complex plane. The latter then satisfies, for the purposes of CWT and SHM, continuity and generality properties. The introduced methods are then implemented on experimental data, of a masonry wall and a railway bridge, instrumented by accelerometric sensors.

In the case of structures under ambient loading, two contributions to the use of CWT are proposed. The first one focuses on the autocorrelation estimator, for wavelet modal analysis of structures with linear dynamic behavior. Its statistical error, resulting from its estimation from a limited measurement time and particularly impacting the calculation of damping, is studied analytically, numerically and experimentally. The second one is an original method of stochastic characterization of non-linear and non-stationary systems under ambient excitation using CWT, which provides laws linking the amplitude and frequency of vibration of these systems. The latter is then implemented on signals from numerical simulations of a bridge, damaged by breathing cracks and subjected to road traffic, and then on experimental data from prestressed road bridges under ambient traffic.
This research responds to the need for increasing maintenance of aging structures and infrastructures in France, with the challenge of creating optimized maintenance systems in increasingly constrained budgetary contexts.

The Structural Health Monitoring methods studied in this research is based on dynamic response and therefore the monitoring of the structural vibration behavior.

This global information is major to complete the measurements deployed on infrastructures because it makes it possible to detect and then locate a non-visible defect by inspection, for example a beginning of internal damage or a lack of support.

Vibratory monitoring of civil engineering works aims at detecting an early structural failure with the early detection of damage. In most cases, a vibration study can be performed in service condition, under the effect of the traffic for example or as a response to external stress, as a strong wind for example.

The difficulty of the diagnosis is then to detect damage (loss of prestressing, failure of a support, cracking of the concrete or corrosion of rebars) and to separate the effects related to the damage of those induced by the external conditions, temperature, sunshine, and hygrometry.

The composition of the civil engineering works, with materials of diverse and heterogenous nature does not facilitate the analysis of data. Indeed, the reinforcements, the prestressing cables and the concrete can all carry different pathologies leading to more or less important disorders, with a competition of the phenomena related to the loss of mass and the inertia. The type of loading can also change the vibratory response.

In this subject, we will first try to detect concrete defects by creating free surfaces and thus predict the development of cracks. The detection of a new event in the vibratory response of a structure under surveillance, or the detection of a difference with a previous state, are particularly suited approaches to signal analysis. Such an approach has a twofold interest to avoid making assumptions a priori on the dynamic behavior and to integrate all the factors of variability.

A particular effort will then be to choose for the studied structure, a model "sufficient" to interpret the modifications of the modes of vibration; for example, representation of free surfaces created to describe cracking defects. In this study, two methods of vibration response processing will be evaluated, a methodology for identifying the vibration behavior of the bridge, the stiffness and damping values evaluated and their evolutions over time measured and the detection of defects optimized.

Deployment of these wavelet analysis methods to a civil engineering structure should be considered. A first real-life test was carried out as part of Nicolas Vacca's thesis on the Jules Verne bridge in Amiens and will be analyzed at the beginning of the thesis. Other complementary laboratory research activities are aimed at defining indicator processing solutions and proposing maintenance strategies. This subject will therefore integrate naturally into the expertise of the laboratory. It will complement the I4S (COSYS / SII) approaches on the development of Structural Health Monitoring (SHM) systems through the coupling of statistical data and physical modeling, bringing a strong application to aging reinforced concrete structures.

Keywords : civil engineering structures, defects, vibratory, wavelet transform

Main host Laboratory - Referent Advisor	-
Director of the main host Laboratory	-
PhD Speciality	Mécanique
Axis of the performance contract	2 - COP2017 - More efficient and resilient infrastructure
Main location	Marne-la-Vallée
Doctoral affiliation	UNIVERSITE PARIS-EST
PhD school	SCIENCES, INGENIERIE ET ENVIRONNEMENT (SIE)
Planned PhD supervisor	ARGOUL Pierre - Université Gustave Eiffel - LVMT
Planned financing	Contrat doctoral - Ifsttar
Abstract To meet the increased need for monitoring aging structures and infrastructures, the study of their vibratory behavior in operational conditions, as part of Structural Health Monitoring (SHM), offers a cheap and promising means of damage detection. For this purpose, the Continuous Wavelet Transform (CWT) is a suitable tool, both in the study of the evolution of modal parameters, and by its ability to detect and characterize non-linear and non-stationary dynamic behaviors. The aim of this thesis is to contribute to the improvement of current analysis techniques based on this transform, and to the development of new methods. In the case of transient or harmonic responses, existing solutions of modal analysis using wavelets are poorly adapted to multi-channel signals. A joint transform for these signals is therefore proposed, allowing a reduction of the relative noise and the definition of a single ridge, along with a criterion of normalization of the instantaneous mode shapes in the complex plane. The latter then satisfies, for the purposes of CWT and SHM, continuity and generality properties. The introduced methods are then implemented on experimental data, of a masonry wall and a railway bridge, instrumented by accelerometric sensors. In the case of structures under ambient loading, two contributions to the use of CWT are proposed. The first one focuses on the autocorrelation estimator, for wavelet modal analysis of structures with linear dynamic behavior. Its statistical error, resulting from its estimation from a limited measurement time and particularly impacting the calculation of damping, is studied analytically, numerically and experimentally. The second one is an original method of stochastic characterization of non-linear and non-stationary systems under ambient excitation using CWT, which provides laws linking the amplitude and frequency of vibration of these systems. The latter is then implemented on signals from numerical simulations of a bridge, damaged by breathing cracks and subjected to road traffic, and then on experimental data from prestressed road bridges under ambient traffic. This research responds to the need for increasing maintenance of aging structures and infrastructures in France, with the challenge of creating optimized maintenance systems in increasingly constrained budgetary contexts. The Structural Health Monitoring methods studied in this research is based on dynamic response and therefore the monitoring of the structural vibration behavior. This global information is major to complete the measurements deployed on infrastructures because it makes it possible to detect and then locate a non-visible defect by inspection, for example a beginning of internal damage or a lack of support. Vibratory monitoring of civil engineering works aims at detecting an early structural failure with the early detection of damage. In most cases, a vibration study can be performed in service condition, under the effect of the traffic for example or as a response to external stress, as a strong wind for example. The difficulty of the diagnosis is then to detect damage (loss of prestressing, failure of a support, cracking of the concrete or corrosion of rebars) and to separate the effects related to the damage of those induced by the external conditions, temperature, sunshine, and hygrometry. The composition of the civil engineering works, with materials of diverse and heterogenous nature does not facilitate the analysis of data. Indeed, the reinforcements, the prestressing cables and the concrete can all carry different pathologies leading to more or less important disorders, with a competition of the phenomena related to the loss of mass and the inertia. The type of loading can also change the vibratory response. In this subject, we will first try to detect concrete defects by creating free surfaces and thus predict the development of cracks. The detection of a new event in the vibratory response of a structure under surveillance, or the detection of a difference with a previous state, are particularly suited approaches to signal analysis. Such an approach has a twofold interest to avoid making assumptions a priori on the dynamic behavior and to integrate all the factors of variability. A particular effort will then be to choose for the studied structure, a model "sufficient" to interpret the modifications of the modes of vibration; for example, representation of free surfaces created to describe cracking defects. In this study, two methods of vibration response processing will be evaluated, a methodology for identifying the vibration behavior of the bridge, the stiffness and damping values evaluated and their evolutions over time measured and the detection of defects optimized. Deployment of these wavelet analysis methods to a civil engineering structure should be considered. A first real-life test was carried out as part of Nicolas Vacca's thesis on the Jules Verne bridge in Amiens and will be analyzed at the beginning of the thesis. Other complementary laboratory research activities are aimed at defining indicator processing solutions and proposing maintenance strategies. This subject will therefore integrate naturally into the expertise of the laboratory. It will complement the I4S (COSYS / SII) approaches on the development of Structural Health Monitoring (SHM) systems through the coupling of statistical data and physical modeling, bringing a strong application to aging reinforced concrete structures.
Keywords :	civil engineering structures, defects, vibratory, wavelet transform

List of topics

Applications closed