Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/35470
Title: SHM-based seismic progressive collapse analysis of RC bridge structures
Authors: Zheng, Yue
Advisors: Xu, You-lin (CEE)
Keywords: Bridges -- Earthquake effects.
Earthquake engineering.
Structural health monitoring.
Issue Date: 2015
Publisher: The Hong Kong Polytechnic University
Abstract: As one of the most damaging natural disasters, strong earthquakes often cause numerous structures to collapse and many people to die, which were reflected over again in recent Wenchuan Earthquake in 2008, Tohoku Earthquake in 2011, and Nepal Earthquake in 2015. To prevent buildings from collapse when they are subjected to strong earthquakes, the dynamic collapse of reinforced concrete (RC) building structures has been investigated actively and extensively, in which the finite element (FE)-method-based collapse analysis of RC building structures is an effective method. However, all the current FE-method-based seismic collapse analyses are based on the removal of entire element, but the actual seismic collapse of a structure often starts from the failure of an element at some degree-of-freedoms (DOFs). The removal of an element without the failure at all its DOFs may lead to false structural collapse. More importantly, the current FE-method-based collapse analysis methods contain many uncertainties. The time-dependent compressive strength of concrete, the confinement effect of core concrete due to the stirrup, the creeping and shrinking effects of concrete, the strength enhancement effect of the reinforcement embedding in the concrete, and the existing damage in concrete and reinforcement due to previous earthquakes, among others, cannot be fully or partially considered in the current collapse analyses, which again may produce false structural collapse. On the other hand, as a cutting-edge technology, structure health monitoring (SHM) systems have been installed in some important buildings and bridges to monitor their functionality, safety, and integrity with the ultimate goal of preventing the buildings and bridges from collapse. Nevertheless, there are seldom studies on how to utilize the information recorded by a SHM system to eliminate the uncertainties existing in the current collapse analysis method and provide an evolutionary and accurate collapse analysis method including collapse prognosis. In view of the problems outlined above, this thesis aims at developing a SHM-based seismic collapse analysis (prognosis) method for RC structures under earthquake excitation. In consideration that the existing studies on RC bridge structures are much less than RC building structures and the collapse mechanism of RC bridge structures may be very different from that of RC building structures, this thesis focuses on seismic collapse analysis of RC bridge structures. A refined collapse analysis method for RC structures considering the DOF release other than the element removal is first proposed in this thesis. By considering the DOF release, the catenary effect of RC beams and the effect of axial force on RC columns can be considered. Three numerical case studies were performed to examine the feasibility and accuracy of the proposed DOF release method. The numerical results of the collapse analysis of a two-span RC continuous beam with its two ends fixed under a concentrated static load was first compared with the experimental results. The result comparisons show that the refined method gives a better agreement with the experimental results compared with the traditional element removal method. The refined dynamic collapse analysis method was then applied to a two-story RC frame structure to demonstrate the entire progress of dynamic collapse. The numerical results demonstrate again that the refined method gives more reasonable collapse results by taking the catenary effect into account than the traditional element removal method. Finally, a two-span continuous RC bridge with a two-column pier at its middle was taken as an example to demonstrate the applicability of the refined method to RC bridge structures. The results show that the collapse of the RC columns does not occur immediately after the DOFs associated with bending moment and shear force of the two columns are released, and that the final collapse of the two columns is due to excessive axial loads. This failure mode could not be predicted by the traditional element removal method. Therefore, the refined method based on DOF release is preferable for the collapse analysis of RC structures including RC bridge structures. A 1:12 scaled RC cable-stayed bridge model was then elaborately designed and constructed to experimentally study the seismic collapse of the RC bridge that was not designed for the seismic resistance and to provide measurement data for implementing the proposed SHM-based collapse analysis method. A comprehensive measurement (SHM) system was designed and installed on the RC bridge to record both the global responses and local responses of the bridge. Before the shaking table tests, each cable force of the as-built RC bridge was measured by the frequency method to ensure that the bridge configuration meets the design requirement. The dynamic characteristic test was then conducted to gain an insight of the properties of the bridge. Finally, a series of earthquake tests, which include small earthquake, moderate earthquake, large earthquake and collapse earthquake in terms of their peak ground accelerations (PGA) and spectra, were conducted. It was observed from the four shake table tests that: (1) the RC bridge performed linearly and elastically under the small earthquake excitation and the RC bridge kept intact conditions after the small earthquake excitation; (2) the RC bridge performed slightly nonlinearly and plastically under the moderate earthquake excitation. The concrete in the failure-vulnerable components cracked slightly; (3) the RC bridge performed severely nonlinearly and plastically under the large earthquake excitation. The concrete in the failure-vulnerable components crushed severely and the reinforcement in the failure-vulnerable components yielded severely; and (4) the RC bridge partially collapsed under the collapse earthquake excitation. The concrete in the failure-vulnerable components crushed severely and the reinforcement in the failure-vulnerable components yielded severely. The measured data acquired from the SHM system together with the dynamic characteristics provide plentiful information for the subsequent linear model updating of the intact bridge, the nonlinear model updating of the damaged bridge, and collapse prognosis for the damaged bridge, respectively. A 3-D FE model of the physical RC cable-stayed bridge subject to shake table tests is established for conducting seismic collapse analysis. To get an accurate FE model of the bridge for further collapse prognosis, a linear model updating strategy using two types of objective functions, the objective function based on natural frequencies and the objective function based on acceleration and strain responses, is proposed with the purpose of updating key parameters of the intact bridge so as to eliminate the uncertainties related to the linear RC bridge structure. A total of 12 key parameters are identified by virtue of sensitivity-based FE model analyses and updated using natural frequencies as the objective function. Three accelerometers and three strain gauges are selected as the key sensor locations and their responses are used for the further model updating in the time domain. Various seismic response time histories computed using the two different updating methods are compared with the measured responses. The comparison results indicate that the two objective functions both can improve the quality of the FE model. The second objective function not only can be used as an alternate of the first one for nonlinear model updating but also provides better updating results than the first objective function.
A nonlinear model updating method by using the measured acceleration responses and reinforcement strains of the RC cable-stayed bridge in the time domain is proposed to update the envelop curves of the materials of the nonlinear bridge without knowing its previous loading history. In the nonlinear model updating, the degradations of both unloading stiffness and reloading stiffness are accomplished in addition to the strength degradation. A total of 58 key parameters divided into the five groups are introduced to be updated. The optimization objective function used in the time domain is the same as the one presented for the linear model updating. Three accelerometers and three strain gauges are selected as the key sensor locations and their responses are used for the nonlinear model updating. The updated 58 key parameters are used to configure the envelop curves of the materials of the bridge due to the previous earthquakes and these curves are then used to calculate the seismic responses of the bridge subject to current earthquake excitation. Various seismic response time histories computed using the nonlinear updated results are compared with the measured responses. The comparison results indicate that the updated results of the key parameters are correct and the nonlinear model updating method is feasible. To further verify the SHM-based seismic collapse prognosis method, it is applied to the RC bridge subject to the collapse test. Since the seismic collapse prognosis of a structure shall be carried out based on the current damage conditions of the structure, the SHM-based nonlinear model updating is necessary to find out the current damage conditions of the structure. In this regard, the 58 key parameters of the RC bridge were updated by considering the bridge subject to the latest earthquake ground motion and using the nonlinear model updating method proposed. The results show that the values of the most updated parameters of the bridge under the latest large earthquake excitation became much smaller compared with those identified for the bridge subject to a moderate earthquake excitation. The values/thresholds of the failure criteria of the four zero-length failure elements of the RC bridge were also determined based on the current damaged conditions and compared with those from the undamaged conditions. The comparative results show that the values/ thresholds of the failure criteria of the four zero-length failure elements of the RC bridge determined based on the current damaged conditions are very different from those based on the undamaged conditions. The collapse prognosis of the RC bridge subject to two future earthquake ground excitations were finally performed base on the updated FE model of the bridge to find out which earthquake will cause the true bridge collapse. The computed results showed that the RC bridge did not collapse when it was subjected to the first future earthquake excitation of relatively small intensity. The computed results showed that when the bridge was subjected to the second earthquake excitation of relatively large intensity, the RC southwest pier, as one of the failure-vulnerable component, triggered the flexure failure at 3.133 second of the earthquake excitation and it was separated from the RC bridge structure. The other three failure-vulnerable components experienced severe damage but not failed. A series of computed seismic responses such as acceleration, strain and reaction force of the RC bridge subject to the second earthquake excitation were compared with the shake table test results recorded by the SHM system installed on the bridge. The comparison results showed that the computed results and collapse process are compatible with the test results recorded by the SHM system. The SHM-based collapse prognosis proposed in this chapter is feasible and effective. An ideal SHM system installed on a bridge is useful in monitoring the loading conditions, updating the FE model, assessing the linear and nonlinear performance, and making collapse prognosis of the bridge. Two sets of SHM systems for the prototype RC bridge are established using two different methods. These systems have demonstrated significantly different results in terms of sensor location and sensor number. The SHM system that uses the multi-sensor placement method includes 16 strain gauges and 12 accelerometers, whereas the proposed SHM system includes 24 strain gauges and 10 accelerometers. The sensors of the strain gauge in the proposed SHM system are all placed in the failure-vulnerable components, such as the tower legs beneath the girder and the two south piers, whereas only several strain gauges in the current SHM system are placed on the girder, transverse beam, and stay cable. These differences can be attributed to the fact that the current SHM system is utilized to assess the linear performance of the bridge under service loadings, whereas the proposed SHM system is used to make a collapse prognosis of the RC bridge under seismic loadings. To demonstrate the difference of the current collapse analysis method and the proposed SHM-based collapse analysis method, an SHM-based collapse prognosis of the earthquake-damaged bridge was conducted. In reality, the damage condition of the bridge can be determined using the nonlinear model updating technique in Chapter 6 and the information that is acquired from the proposed SHM system installed on the bridge. In this way, the uncertainties in the FE model of the bridge can be eliminated and provide a critical support for the collapse prognosis. In this study, the damaged state was specified by conducting a nonlinear seismic analysis for absence of SHM system. For comparison, a collapse analysis of the prototype RC bridge is also conducted using the current collapse analysis method. The entire collapse processes from the two collapse analysis/prognosis methods are significantly different: (1) the seismic intensity used for the current collapse analysis method is larger than that used for the proposed SHM-based collapse prognosis method; and (2) the proposed method has detected a partial collapse in the southwest pier, whereas the current collapse analysis method has detected a partial collapse in the southeast pier. Therefore, the proposed SHM-based collapse prognosis is a promising method to prognosticate the behavior of the existing RC bridges under future earthquakes, whereas the current collapse analysis can be only used for the bridge at design stage.
Description: PolyU Library Call No.: [THS] LG51 .H577P CEE 2015 Zheng
xxiv, 324 pages :illustrations
URI: http://hdl.handle.net/10397/35470
Rights: All rights reserved.
Appears in Collections:Thesis

Files in This Item:
File Description SizeFormat 
b28238072_link.htmFor PolyU Users203 BHTMLView/Open
b28238072_ir.pdfFor All Users (Non-printable)7.51 MBAdobe PDFView/Open
Show full item record

Page view(s)

115
Last Week
4
Last month
Checked on Aug 13, 2017

Download(s)

57
Checked on Aug 13, 2017

Google ScholarTM

Check



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