Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/86324
Title: Multi-sensing and multi-scale monitoring of long-span suspension bridges
Authors: Zhang, Xiao-hua
Degree: Ph.D.
Issue Date: 2013
Abstract: An in-depth understanding of the structural response and behavior of a long-span suspension bridge at various levels is required to efficiently reach the objectives of structural health monitoring (SHM). Therefore, multi-scale response reconstruction at key structural locations where sensors are not available is essential. This thesis is devoted to the determination of the number and optimal locations of multi-type sensors and the reconstruction of multi-scale responses of long-span suspension bridges to provide comprehensive information for multi-scale structural monitoring. Strain response is often used to monitor local responses of a structure, whereas displacement is often applied to assess the serviceability and integrity of a structure. This thesis first proposed a dual-type sensor optimal placement and multi-scale response reconstruction method. The number and locations of dual-type sensors, which consist of strain and displacement measurement sensors, are optimized to obtain the best reconstruction of multi-scale structural responses, which are strain (local) and displacement (global) responses. The locations of the two types of sensors are selected simultaneously, and the total number of sensors is determined by a threshold of the total response reconstruction error. A numerical case study of a cantilever beam is presented as an example to demonstrate the feasibility of the proposed method. The numerical case study demonstrates the feasibility of the proposed method. The dual-type sensor system can provide more comprehensive and accurate information for multi-scale response reconstruction than single-type sensor systems. Accelerometers, which are widely used in civil engineering because of their convenience and sensitivity, are not considered at this stage. The multi-type sensor optimal placement and multi-scale response reconstruction method are further developed to include accelerometers in addition to strain and displacement measurement sensors. Accelerometers are also used for measuring global structural response. The number and locations of multi-type sensors are selected simultaneously through the optimization procedure to minimize the reconstructed errors between real and reconstructed responses. The Kalman filter algorithm is employed in this method. A simply supported overhanging steel beam and a 25-bar truss structure are analyzed as numerical examples to examine the effectiveness of the proposed method. The results of the numerical studies indicate that the reconstructed responses satisfactorily match the real responses. The reconstructed multi-scale responses using multi-type measurements from optimally selected locations are more accurate than the responses obtained using measurements from single-type sensors or from similar but differently placed multi-type sensors. Experimental investigations of the proposed dual- or multi-type sensor location optimization and multi-scale response reconstruction methods using a simply supported overhanging steel beam are then conducted before the application of the two methods to complex structures. Two experiments are conducted separately to investigate the proposed methods. The first experiment investigates dual-type sensor optimal placement and multi-scale response reconstruction method. The second investigates multi-type sensor optimal placement and multi-scale response reconstruction method. Multi-scale structural responses at key locations are reconstructed based on data fusion of measurement results from selected locations using the proposed methods, including local and global responses. The results are compared with a number of redundant measurements to assess the accuracy of the reconstructed responses. All experimental investigation results indicate that dual- or multi-type measurements from optimally selected locations can effectively reconstruct multi-scale responses with reasonable accuracy, thus providing comprehensive global and local information for SHM.
At present, applying the proposed methods directly to real long-span suspension bridges is extremely challenging because of the large size and complex structure of these bridges. A laboratory-based testbed is established to overcome the aforementioned difficulties in multi-scale monitoring of long-span suspension bridges. The design principles of the testbed and the details of the design and setup of the physical bridge model are presented. The geometric configuration and dynamic characteristics of the testbed are identified through geometric measurements and modal tests. The corresponding finite-element (FE) modeling of the physical bridge model is also established using a commercial software package followed by FE model updating in terms of measured modal properties. The testbed comprising of the physical model and the updated FE model can serve as a benchmark problem for the SHM of long-span suspension bridges, including testing various methods for structural monitoring. The testbed is then used numerically and experimentally to extend the application of the multi-type sensor optimal placement and multi-scale response reconstruction method. The procedure for multi-type sensor location selection for the physical bridge model subject to ground motion differs to some extent from the procedure for a simple structure because of the complexity of the structure and modes of the model. A framework for multi-scale monitoring of the testbed is presented, and a practical method is proposed to select the modes through the reduction of the relative percentage errors (RPEs) between real and reconstructed responses for sensor location selection and response reconstruction. Numerical studies and experimental investigations are then carried out. Investigation results indicate that the reconstructed responses using the measurements from the sensors at the selected locations agree well with real or measured responses. The testbed is also used experimentally and numerically to investigate the effects of support settlement and cable slippage. Dead loading tests are conducted to examine the integrity of the physical bridge model, to check the functionality of the sensing system, and to verify the FE model. Four settlement tests are subsequently studied experimentally, including two tests for anchorage settlements and two tests for tower settlements, to investigate the static behavior caused by the support settlement. In addition, four cable slippage tests are experimentally investigated. Cable slippage tests are performed on the saddles of the two towers in two directions. The counterpart settlement and slippage simulation using the FE model of the testbed is also performed, and the simulation results are compared with those from the tests. Finally, numerical studies are conducted using the FE model of the testbed to investigate cable slippage caused by a large support settlement. Investigation results show that the tests results agree well with the computed values from the FE model of the testbed. The effect of the support settlement on tower strains, as well as on cable forces and strains, is local. Nevertheless, the support settlement affects deck strains and vertical displacements through the whole deck system. Cable slippage has little effect on the deck regardless of which saddles and directions the cable slips. However, cable slippage greatly affects tower strains, cable forces and strains near the span where the cables slip. Slippage may occur if the structure is subject to abnormal external loading.
Subjects: Suspension bridges -- Testing.
Structural health monitoring.
Hong Kong Polytechnic University -- Dissertations
Award: FCE Awards for Outstanding PhD Theses
Pages: xxxiii, 295 p. : ill. ; 30 cm.
Appears in Collections:Thesis

Show full item record

Page views

54
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.