Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/85111
Title: SHM-based condition assessment of in-service bridge structures using strain measurement
Authors: Xia, Hongwen
Degree: Ph.D.
Issue Date: 2012
Abstract: In realistic scenarios, in-service bridge structures are at risk from structural degradation, service demands of increasing traffic flow and heavier truck loads, natural or man-made disasters, or deferred maintenance. Condition assessment of these public facilities for future serviceability and safety is a challenging task to their owners and engineers. A paradigm of integrating structural health monitoring (SHM) data into procedures of structural condition assessment is expected to achieve objective and quantitative condition assessment in practice. The work addressed in this dissertation has been dedicated to investigating condition assessment of existing bridge structures using strain measurement acquired under in-service environment by a long-term bridge health monitoring system (BHMS). Under in-service circumstance, bridge structures are subject to temperature variation, traffic and wind effect, and material deterioration due to aging or aggressive environmental attack. Strain measurement acquired under operational environment by the SHM system is naturally a result of combination of these external loadings and environmental effects. Source separation of these effects from the raw measurement is a challenging job and it is pursued in this study. The proposed method takes the benefits of discrete wavelet transform (DWT) which satisfies the mathematical principle of multi-component separation quite well (less distortion and cross-talking among components). Based on the wonderful decomposition platform of DWT, component extraction with physical meanings is realized by integrating physical mechanism of the desired structural behavior into the selection criterion of source separation. Specific application of the proposed method in component analysis of the measured strains from the Tsing Ma Bridge (TMB) deck exemplifies its effectiveness in source separation of multi-component strain monitoring signals. By interpreting separated strain components from the raw measurement, structural behaviors of the TMB deck under temperature effect, live load effect and traffic effect are identified for further condition assessment. When using the strain response data caused by live load effect for reliability assessment, another problem arises in the inference of probability distribution from the observed data that strain response due to live load effect collected under in-service environment is still a result of multi-load effect such as traffic (highway, railway, or both of them) and wind (monsoon or typhoon), and it cannot be characterized by a standard probability distribution model adequately. Mixed distributions existing in the monitoring data are explored primarily by histogram analysis. Then hybrid mixture estimation including model selection and parameter estimation is evaluated and a structure of mixed Weibull model is proposed for the probability density function (PDF) estimation of peak-stress values counted from the derived stress-time histories. Different mixture models (e.g., normal mixtures, log-normal mixtures and Weibull mixtures) are compared in the process of model selection by calculating Akaike's information criterion (AIC) values. Convergences of AIC value with a varying component number are addressed for the optimal determination of component numbers in mixture models. Based on the estimated PDFs, reliability based condition assessment of the performance of the TMB deck trusses is carried out for various load scenarios such as monsoon, typhoon, with and without railway traffic.
As another target pursued in this study, strain response data due to traffic effect are used to estimate the neutral-axis position of the monitored deck section for further application in damage detection. It is revealed by the monitoring data that under traffic effect the TMB deck performs as a flexural beam, i.e., the deck top compresses and the deck bottom tensions concurrently, or vice versa. Based on the relation between the neutral-axis position and strain responses at the top and bottom of a cross section, a Kalman filter (KF) estimator for locating the neutral-axis position from strain measurement is elaborately designed and in succession comes the validation of its stability to noise disturbance through numerical studies. Two levels of noise contamination (5% and 10%) in the sensor readings are considered in the simulation studies. Results of the numerical simulation show that the KF based estimation method can generate better results in comparison with a direct estimation approach. Moreover, application of the proposed KF estimator to the neutral-axis position estimation of the instrumented TMB deck sections demonstrates its efficiency in the real monitoring data. To further testify the feasibility of neutral-axis position as a damage indicator, experiment and numerical simulation are conducted to demonstrate its sensitivity to damage and independence of traffic load patterns. In design of the experiment, a flexible steel beam subject to moving bogies is objectively used as the testing model to simulate the structural behavior of a bridge deck under the traffic effect and a cut on a selected cross section is to simulate damage incurred on the testing model. By establishing a multi-scale finite element method (FEM) model, numerical simulation for damage detection (using the neutral-axis position as an indicator) of the testing model is conducted for static and moving load cases respectively. Physical experiment on damage detection of the testing model follows and the previously designed KF estimator is used to locate the neutral-axis position from the noisy experiment data. Results of the numerical simulation and physical experiment show that the neutral-axis position can serve as a good indicator of damage and it can be conveniently achieved through strain monitoring in practice. In summary, the research described in this dissertation chiefly contributes to the development of a systematic framework for condition assessment of existing bridge structures making use of long-term monitoring data of strain responses. This approach involves the multi-component analysis of strain monitoring signals, mixture distribution model based reliability assessment, Kalman filter based optimal estimation of neutral-axis position, and neutral-axis position based damage detection. Following this approach, objective and quantitative condition assessment of in-service bridge structures can be achieved with the use of SHM data.
Subjects: Bridges -- Inspection.
Strains and stresses -- Measurement.
Structural health monitoring.
Hong Kong Polytechnic University -- Dissertations
Pages: xxvi, 200 leaves : ill. (some col.) ; 30 cm.
Appears in Collections:Thesis

Show full item record

Page views

68
Last Week
0
Last month
Citations as of Mar 24, 2024

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.