Dominant Failure Mode-based Structural Damage Identification Methods

For significant infrastructure facilities,such as long-span bridges,ensuring their safety in service is a key mission for civil engineering community.It is well known that the structural health monitoring(SHM)technology is of great potential to achieve such goal.Damage identification(DI)is a key to success in SHM applications.Therefore,the DI research has attracted a lot of attention for decades.However,there are still vast needs for studies in the field.Hence,this dissertation focuses on the innovation on structural damage detection research methodology by introducing a concept,dominant failure modes,which comes from the structural reliability theory,and suggests an idea,developing damage identification methods based on characteristics occurring during the failure of structures.Usually,most damage identification techniques are state comparison-based,i.e.a SHM system detects the presence,location and severity of structural damage by comparing two different structural states.The undamaged state is commonly selected as the reference.Both theoretical analysis and practical applications have shown that the effectiveness of most established methods is limited and innovation is urgently needed.This study shifts the reference from the undamaged state to the failure state of a specific dominant failure mode,and introduces a new structural damage identification research frame by highlighting dominant failure modes.New novel DI approaches are proposed by fully utilising unique features and dynamical properties of structures failed in these modes.The main contents of this dissertation are as follows:(1)To identify damage related to specific dominant failure modes,first of all,the issues of searching for these modes and estimating their occurrence probability are investigated.Firstly,the concept of dominant failure modes is introduced and common techniques employed to search the modes are briefed.Secondly,to analyse the dominant mode of simply supported beams--the flexural mode,a theoretical method and failure probability estimation algorithm are discussed.A unique,interesting characteristic,linearized deflection curves,observed on beams failed in the mode is highlighted.For fixed arch structures,a method adopted to search for one of their dominant modes,the mechanism failure mode,is developed.Finally,the distinction between the damage failure mode-specific research strategy and common studying methodology is investigated,and then a new damage identification paradigm is proposed.(2)In order to verify the newly-developed method presented previously,a numerical simulation and an arch experiment are carried out,and the failure process and paths of the specimen are examined.A finite element model(FEM),using frame elements,of a masonry fixed arch is built and then utilizing the method searches for all failure paths.A concrete block arch specimen is manufactured and then loaded to crash.The forming sequence of all hinges during the failure is analysed.The results show that developing five hinges leads to the structural failure;that the appearing order of the hinges and the failure procedure both match very accurately with those predicted in the theoretical analysis and the FEM simulation.(3)Based on the evolving paths of the natural frequencies and modal shapes occurring when fixed arches fail in the mechanism mode,new vibration-based damage identification theory and methods are developed.Firstly,according to the analysis on the equation of motion(EOM),the fact,part of the fundamental factors of structural vibration-based damage detection methods are the essential attributes of structures,such as stiffness,is detailed.The correspondence among dynamic response variations,the structural nature variations and different stages of a specific failure mode is analysed,then the following two-step research idea is proposed;the first step is to obtain the changing paths of the structural stiffness matrix and dynamic properties by analysing the failure process of the structures following the mode;the second step is to develop corresponding damage identification methods by characterizing the evolution of the dynamic properties.Secondly,based on the frequency and the modal shape variations obtained on fixed arches failed in the mechanism mode,two methods are developed.The first one is perturbation method-based and the other diagnoses damage based on changes in multiple frequencies and mode shapes.These methods are numerically verified with the observations and data from the arch test detailed in Chapter 3.(4)For the flexural failure mode of simply supported beams,a new damage identification method and theory are proposed by quantitatively describing a unique characteristic occurring during the failure of the beams in the mode.Firstly,a very interesting characteristic,the deflection curve linearization,is observed on failing beams.Secondly,an analytical model is built to link the shapes of a residual beam segment to the state of structural damage and,therefore,theoretically connect the degree of the deflection curve linearization with damage severity.Furthermore,the quantification of the damage severity is transformed to a problem,characterizing the difference between a deflection curve and its secant(termed the limit line).Finally,the Fourier series decomposition-based damage identification method is presented.Namely,the functions of both the deflection curve and its limit line of a beam(or a segment)are expressed in a Fourier series form,which produces two sets of Fourier coefficients,and then the damage severity is quantified with the angle between the two vectors whose elements come from the two sets.The damage is located precisely by searching for the position on a deflection curve where the first-order derivative is discontinuous.(5)To verify the Fourier series decomposition-based damage identification method,an experiment on reinforced concrete(RC)beams is conducted.Eight RC beams are manufactured and then loaded to failure.The results show that the linearized deflection curve characteristic occurs on all specimens and eventually each beam fails due to the formation of a hinge.The presence,location and severity of beam damage are identified exactly by adopting the newly-developed method.This investigation proves that the damage on all specimens can be detected precisely with the Fourier series decomposition-based method.Based on the aforementioned work,this dissertation has confirmed the feasibility and the effectiveness of the research strategy,conducting damage identification research specific to dominant failure modes.