Output-only Damage Detection Of Bridge Structures Based On Vibration Data And Experimental Study

Infrastructures,such as bridges,play an important role in city life and receive various kinds of excitation during their operational life.Structural Health Monitoring(SHM)is a tool to control safety and detect possible damage from structural responses.In some cases,it is impossible to measure the operational load exerted on a bridge,e.g.under heavy traffic highway.In such cases,where a complete knowledge of excitation is not available,output-only modal identification and damage detection methods can be used.This thesis develops three output-only signal-processing based damage detection methods to locate damage from acceleration data.These acceleration data can be obtained from either numerical model or experimental works.Utilizing acceleration data to detect the damage location could effectively decrease the cost of SHM.Accelerometers can be implemented in any desired location of structure to record the acceleration response with a wide range of frequencies.The first method employs Moving Average Filter(MAF)to develop an output-only baseline-free damage detection algorithm to locate damage in a steel beam under moving load.MAF is a convolution approach based on a simple filter kernel(rectangular shape)that works as an averaging method to smooth signal and remove incorporated noise.To verify the first method,a simply supported beam was 3D numerically simulated and experimentally modelled under a moving vehicle load.Damage in these 3D numerical and experimental models was artificially made as a crack with width 3 mm at the bottom of flanges.A bogie was also designed and constructed to simulate the moving load with constant speed.All acceleration signals were recorded at nine stations from either numerical model or experimental works simultaneously with a sampling frequency of 500 Hz.The results prove that the smoothed signals can show an abnormal vibration close to the damage location as the truckload passes the damage location.The first method can accurately estimate the damage location by only one accelerometer in a noise-free environment or a set of accelerometers along the beam in the laboratory model(noisy environment).The results prove that the first method accurately determines the damage location in noisy signals with no need to use any filtering or baseline correction methods.The second method uses Random Decrement(RD)technique,introduced by Cole in 1968,incorporating with Arias intensity(AI)to locate the damage place from acceleration data recorded on a simply supported beam and a tied-arch bridge.RD technique is an output-only averaging technique that is used in cases where the control or initial excitation cannot be measured.Since the dynamic response of structure contains transient response and free response alike,RD technique averages the transient response out;what remained called RD functions that behave similar to free response.It was proved that these RD functions have valuable information about modal parameters and possible damage.To locate the damage in the second method,AI is employed to calculate the energy content of each RD function and substitute each acceleration signal by a scalar value.Normalizing AIs,all RD functions are then updated so as to show a unique energy along the undamaged structure.Once the normalizing factor are computed for the intact structure,the damage are determined by the absolute difference of normalized AIs obtained from each individual RD function along the structure simultaneously.To verify the second method,a simply supported beam(the laboratory model used to verify the first method)and a tied-arch bridge were experimentally modelled under a moving vehicle load with constant speed.Damage in these two models was artificially made as a section area attenuation in the cable and the beam flanges.For the tied-arch bridge,a bogie was built to simulate the traffic load and carry about 130 kg load with constant speed.All acceleration signals were recorded at certain stations from either the simply supported beam or the tied-arch bridge,simultaneously,with sampling frequencies of 500 and 1000 Hz.It was tried to implement the accelerometers as close as possible to the cable supports on the deck of the tied-arch bridge.The results indicate that the second method could accurately determine the damaged cable and crack in noisy signals with no filtering or baseline correction.The third method uses seismic acceleration response of bridge as input to locate damaged cable in a tied-arch bridge.It is well known that a bridge can receive severe damage under the aftershocks even if it stands the main shock.As such,a quick assessment seems necessary when a bridge experiences an earthquake event.In the third method,seismic acceleration response signals are first processed by a simple band-pass filtering.Their energy level(in terms of Arias intensity)are then calculated and normalized to locate the damaged cables.The third method was verified via a 3D numerical model of a tied-arch bridge under 20 different earthquakes.Damage was modelled as section area attenuation in the cable.All seismic acceleration response signals were recorded at top of the tied-arch bridge's deck close to the cables,simultaneously.The results prove that the third method accurately locates the damaged cable under the earthquake.Additionally,it is shown that the near-and far-field parameters regarding the earthquake have no certain effects on the damage detection procedure.Three different damage indexes(a damage index for each damage detection method)were developed based on the energy of signal.These damage indexes were then adopted to assign a scalar value to each signal.This issue enabled us to represent all acceleration signals in a single bar chart.Having this chart,it is then easy to locate the damage by finding the maximum between the bars calculated from energy-based damage indexes.The three proposed methods could properly find the singularity in acceleration signal with no need to calculate the modal parameters,representing a time-and cost-effective method in structural health monitoring.Furthermore,these three methods do not need to know the place of damage in advance that make them categorized as a forward method.In addition,these methods are formulated as an energy-based solution to locate the damage without interfering the frequency variation.That means,frequency shifting due to the damage presents no certain effects on damage detection procedure.