This research studies finite element (FE) model updating formulations utilizing the measured frequency-domain modal properties, i.e. resonance frequencies and mode shapes. The modal properties provided by an FE model are usually different from the ones experimentally measured from an as-built structure. To update the FE model parameters, optimization problems are formulated to minimize the difference between experimental and simulated modal properties. Two modal property difference formulations are presented in this research, one using MAC values and the other using direct differences between eigenvectors. To find the optimal solution of the formulated optimization problem, two optimization algorithms are studied for comparison, i.e. the Levenberg-Marquardt and the trust-region-reflective algorithms. Randomly generated starting values of optimization variables are adopted to increase the chance of finding global minimum. The model updating formulations with different optimization algorithms are studied with a space frame example.
In order to obtain a finite element (FE) model that can more accurately describe structural behaviors, experimental data
measured from the actual structure can be used to update the FE model. The process is known as FE model updating. In
this paper, a frequency response function (FRF)-based model updating approach is presented. The approach attempts to
minimize the difference between analytical and experimental FRFs, while the experimental FRFs are calculated using
simultaneously measured dynamic excitation and corresponding structural responses. In this study, the FRF-based model
updating method is validated through laboratory experiments on a four-story shear-frame structure. To obtain the
experimental FRFs, shake table tests and impact hammer tests are performed. The FRF-based model updating method is
shown to successfully update the stiffness, mass and damping parameters of the four-story structure, so that the analytical
and experimental FRFs match well with each other.
In structural sensing applications, wireless sensing systems have drawn great interest owing to faster installation process and lower system cost compared to the traditional cabled systems. As a new-generation wireless sensing system, <i>Martlet</i> features high-speed data acquisition and extensible layout, which allows easy interfacing with various types of sensors. This paper presents a field test of the <i>Martlet</i> sensing system installed at an in-service pre-stressed concrete highway bridge on SR113 over Dry Creek in Bartow County, Georgia. Four types of sensors are interfaced with <i>Martlet</i> in this test, including accelerometers, strain gages, strain transducers and magnetostrictive displacement sensors. In addition, thermocouples are used to monitor the temperature change of the bridge through the day. The acceleration, strain and displacement response of the bridge due to traffic and ambient excitations are measured. To obtain the modal properties of the bridge, hammer impact tests are also performed. The results from the field test demonstrate the reliability of the <i>Martlet</i> wireless sensing system. In addition, detailed modal properties of the bridge are extracted from the acceleration data collected in the test.
To improve the simulation accuracy of the finite-element (FE) model of an as-built structure, measurement data from the actual structure can be utilized for updating the model parameters, which is termed as FE model updating. During the past few decades, most efforts on FE model updating intend to update the entire structure model altogether, while using measurement data from sensors installed throughout the structure. When applied on large and complex structural models, the typical model updating approaches may fail due to computational challenges and convergence issues. In order to reduce the computational difficulty, this paper studies a decentralized FE model updating approach that intends to update one substructure at a time. The approach divides the entire structure into a substructure (currently being instrumented and updated) and the residual structure. The Craig-Bampton transform is adopted to condense the overall structural model. The optimization objective is formulated to minimize the modal dynamic residuals from the eigenvalue equations in structural dynamics involving natural frequencies and mode shapes. This paper investigates the effects of different substructure locations and sizes on updating performance. A space frame example, which is based on an actual pedestrian bridge on Georgia Tech campus, is used to study the substructure location and size effects. Keywords: substructure
This research studies a substructure model updating approach. Requiring modal testing data from only part of a large structure (i.e. a substructure), finite element model parameters for the substructure can be updated. Prior to updating, Craig-Bampton transform is adopted to condense the entire structural model into the substructure (currently being instrumented and to be updated) and the residual structure. Finite element model of the substructure remains at high resolution, while dynamic behavior of the residual structure is approximated using only a limited number of dominant mode shapes. To update the condensed structural model, physical parameters in the substructure and modal parameters of the residual structure are chosen as optimization variables; minimization of the modal dynamic residual is chosen as the optimization objective. An iterative linearization procedure is adopted for efficiently solving the optimization problem. The proposed substructure model updating approach is validated through numerical simulation with two plane structures
The introduction of wireless telemetry into the design of monitoring and control systems has been shown to reduce
system costs while simplifying installations. To date, wireless nodes proposed for sensing and actuation in cyberphysical
systems have been designed using microcontrollers with one computational pipeline (i.e., single-core
microcontrollers). While concurrent code execution can be implemented on single-core microcontrollers,
concurrency is emulated by splitting the pipeline’s resources to support multiple threads of code execution. For
many applications, this approach to multi-threading is acceptable in terms of speed and function. However, some
applications such as feedback controls demand deterministic timing of code execution and maximum computational
throughput. For these applications, the adoption of multi-core processor architectures represents one effective
solution. Multi-core microcontrollers have multiple computational pipelines that can execute embedded code in
parallel and can be interrupted independent of one another. In this study, a new wireless platform named Martlet is
introduced with a dual-core microcontroller adopted in its design. The dual-core microcontroller design allows
Martlet to dedicate one core to standard wireless sensor operations while the other core is reserved for embedded
data processing and real-time feedback control law execution. Another distinct feature of Martlet is a standardized
hardware interface that allows specialized daughter boards (termed wing boards) to be interfaced to the Martlet
baseboard. This extensibility opens opportunity to encapsulate specialized sensing and actuation functions in a wing
board without altering the design of Martlet. In addition to describing the design of Martlet, a few example wings
are detailed, along with experiments showing the Martlet’s ability to monitor and control physical systems such as
wind turbines and buildings.