Hamed Ebrahimian

Material Parameter Identification in Distributed Plasticity FE Models of Frame-Type Structures Using Nonlinear Stochastic Filtering

AbstractThis paper proposes a novel framework that combines high-fidelity mechanics-based nonlinear (hysteretic) finite-element (FE) models and a nonlinear stochastic filtering technique, referred to as the unscented Kalman filter, to estimate unknown material parameters in frame-type structures. The proposed identification framework updates nonlinear FE models using spatially limited noisy measurement data, and it can be further used for damage identification purposes. To validate its effectiveness, robustness, and accuracy, this framework is applied to a cantilever steel column representing a bridge pier and two-dimensional steel frame. Both structures are modeled using beam-column elements with distributed plasticity and are subjected to a suite of earthquake ground motions of varying intensity. The results indicate that the material parameters of the nonlinear FE models are accurately estimated provided that the loading intensity is sufficient to exercise the parts (branches) of the nonlinear material...

Shake table testing of a full-scale five-story building: system identification of the five-story test structure

A full scale five-story reinforced concrete building was built and tested on the NEESUCSD shake table. The purpose of this test was to study the response of the structure and nonstructural components and systems (NCSs) and their dynamic interaction during seismic excitations of different intensities. The building specimen was tested under base-isolated and fixed-based conditions. To analyze the effects of the construction process, NCSs, the isolation system and structural/nonstructural damage on the dynamic properties of the building, a sequence of dynamic tests were performed on the test specimen, including ambient vibrations, impact/free vibration and forced vibration tests, the last including low amplitude white noise (WN) and seismic base excitation. In this paper, the measured response data from accelerometers are used to estimate the modal parameters (natural frequencies, damping ratios and mode shapes) of the building at different stages of construction, during the placement of NCSs and at different damage states of the structure and NCSs. Different system identification methods, including output-only and inputoutput, are used for this purpose. The results clearly show the effects of changing the state of the building on its equivalent linear dynamic properties, providing significant insight on the effects, absolute and relative, of the construction activities, NCSs and structural/nonstructural damage on the modal properties of the building.

Shake Table Testing of a Full-Scale Five-Story Building: Pre-test Simulation of the Test Building and Development of an NCS Design Criteria

This paper is one of five within a session discussing the findings from a full-scale, five-story building test program conducted at the NEES-UCSD Large High Performance Outdoor Shake. This paper focuses on the pre-test simulation of the test building with the objective of predicting its seismic response and thereby; (i) guiding the selection of earthquake motions for the test program and (ii) providing guidance for the design of various nonstructural components and systems (NCSs) to be installed in the test building. Two nonlinea,r finite element models were developed independently in this effort: 1) a macro-element based model was implemented in the OpenSees platform, and 2) a detailed finite element model was prepared using the general finite element software package DIANA. In this paper, the description and assumptions adopted for each of these two models are first provided, and then the simulation results for selected test motions are presented and compared. Utilizing these models, a procedure is developed to define project-specific NCSs design criteria. The developed design criteria are compared with the current code provisions and their implications discussed.

Pretest Nonlinear Finite-Element Modeling and Response Simulation of a Full-Scale 5-Story Reinforced Concrete Building Tested on the NEES-UCSD Shake Table

A full-scale five-story reinforced concrete building specimen, outfitted with a variety of nonstructural components and systems (NCSs), was built and tested on the Network for Earthquake Engineering Simulation at the University of California, San Diego (NEES–UCSD) large outdoor shake table in the period March 2011–June 2012. The building specimen was subjected to a sequence of dynamic tests including scaled and unscaled earthquake motions. A detailed three-dimensional nonlinear finite-element (FE) model of the structure was developed and used for pretest response simulations to predict the seismic response of the test specimen and for decision support in defining the seismic test protocol and selecting the instrumentation layout for both the structure and NCSs. This paper introduces the building specimen and the shake table test protocol and describes the techniques used for the nonlinear FE modeling and response simulation. Utilized as blind prediction, the pretest simulation results at different scales (global structural level and local member/section/fiber levels) are compared with their experimental counterparts for seismic input (base excitation) of increasing intensity from serviceability to design levels. The predictive capabilities of the used FE modeling techniques are evaluated and possible sources of discrepancies between the FE predictions and experimental measurements are investigated and discussed.

Pre-Test Nonlinear FE Modeling and Simulation of a Full-Scale Five-Story Reinforced Concrete Building

A full-scale reinforced concrete (R/C) building specimen, furnished with a variety of nonstructural components and systems, was built and tested on the UCSD-NEES outdoor shake table. The building specimen was subjected to a sequence of dynamic tests including scaled and unscaled historical earthquake ground motions. In order to simulate and predict the nonlinear dynamic response of the building specimen, a detailed three-dimensional nonlinear FE model of the structure was developed using the FE analysis software DIANA. By comparing the pre-test simulated results with the experimental results, the effects of the nonstructural components on the dynamic response of the building can be inferred. This paper describes the building test specimen and the nonlinear FE modeling and response simulation. Key numerical results are compared with their experimental counterparts and potential sources of discrepancies are discussed.

Modal Identification of a 5-Story RC Building Tested on the NEES-UCSD Shake Table

A full scale five-story reinforced concrete building was built and tested on the NEES-UCSD shake table. The purpose of this experimental program was to study the response of the structure and nonstructural systems and components (NCSs) and their dynamic interaction during seismic excitation of different intensities. The building specimen was tested under base-isolated and fixed-based conditions. In the fixed-based configuration the building was subjected to a sequence of earthquake motion tests designed to progressively damage the structure. Before and after each seismic test, ambient vibration data were recorded and additionally, low amplitude white noise base excitation tests were conducted at key stages during the test protocol. A quasi-linear response of the building can be assumed due to the low intensity of the excitation and consequently modal parameters might change due to the structural and nonstructural damage. Using the vibration data recorded by 72 accelerometers, three system identification methods, including two output-only (SSI-DATA and NExT-ERA) and one input-output (DSI), are used to estimate the modal properties of the fixed-base structure at different levels of structural and nonstructural damage. Results allow comparison of the identified modal parameters obtained by different methods as well as the performance of these methods and studying the effect of the structural and nonstructural damage on the dynamic parameters. The results show that the modal properties obtained by different methods are in good agreement and that the effect of structural/nonstructural damage is clearly evidenced via the changes induced on the estimated modal parameters of the building.

Evolution of Dynamic Properties of a 5-Story RC Building During Construction

A full scale five-story reinforced concrete building was built and tested on the NEES-UCSD shake table. The purpose of this experimental program was to study the response of the structure and nonstructural systems and components (NCSs) and their dynamic interaction during seismic excitation of different intensities. The building specimen was tested under base-isolated and fixed-based conditions. Furthermore, as the structure was being built, an accelerometer array was deployed in the specimen to study the evolution of its modal parameters during the construction process and due to placement of major NCSs. A sequence of dynamic tests, including daily ambient vibration tests, impact/free vibration and forced vibration (white noise base excitation) tests, were performed on the structure at different stages of construction. Several state-of-the-art system identification methods, including two output-only (SSI-DATA and NExT-ERA) and one input-output (OKID-ERA), were used to estimate the modal properties of the structure (natural frequencies, damping ratios and mode shapes). The results obtained allow to compare the modal parameters obtained from different methods as well as the performance of these methods and to investigate the effects of the construction process and NCSs on the dynamic properties of the building specimen.

Nonlinear Structural Finite Element Model Updating Using Batch Bayesian Estimation

This paper proposes framework for nonlinear finite element (FE) model updating, in which state-of-the-art nonlinear structural FE modeling and analysis techniques are combined with the maximum likelihood estimation (MLE) method to estimate time-invariant parameters governing the nonlinear hysteretic material constitutive models used in the FE model of the structure. Using the MLE as a parameter estimation tool results in a nonlinear optimization problem, which can be efficiently solved using gradient-based optimization algorithms such as the interior-point method. Gradient-based optimization algorithms require the FE response sensitivities with respect to the material parameters to be identified, which are computed accurately and efficiently using the direct differentiation method (DDM). The estimation uncertainties are evaluated based on the Cramer-Rao lower bound (CRLB) theorem by computing the exact Fisher Information matrix using the FE response sensitivities. A proof-of-concept example, consisting of a cantilever steel column representing a bridge pier, is provided to validate the proposed nonlinear FE model updating framework. The simulated responses of this bridge pier to an earthquake ground motion is polluted with artificial output measurement noise and used to estimate the unknown parameters of the material constitutive model. The example illustrates the excellent performance of the proposed parameter estimation framework even in the presence of high measurement noise.

Structural Identification for Dynamic Strain Estimation in Wind Turbine Towers

Fatigue is a common issue in steel structures such as wind turbine towers, which is caused by cyclic wind and wave excitations. Therefore, estimation of the remaining fatigue life of the structural and foundation system is of concern. For this purpose, continuous monitoring of the structure is necessary to obtain strain data at fatigue critical points. Since installing and maintaining strain sensors in critical underwater location is difficult, strain data is often available only from a few sensors at accessible locations. Using these sparse sensors, the strain time histories at fatigue critical points can be estimated using estimation techniques. These techniques can identify the structural system using limited measured response data and a system model. In this paper, we implement a model updating approach followed by modal expansion to estimate the strain time history at critical points in a numerical case study representing an offshore wind turbine tower. The acceleration response of the structure is simulated using a finite element model and polluted with Gaussian white noise to represent measurements. The measurements are then used for model updating and strain estimation. The accuracy of the methods and their robustness to the measurement noise and model uncertainty are investigated. The estimated strain response time histories can later be used as input to an appropriate fatigue damage model to estimate the current state of fatigue damage in the system.

Batch and Recursive Bayesian Estimation Methods for Nonlinear Structural System Identification

This chapter presents a framework for the identification of nonlinear finite element (FE) structural models using Bayesian inference methods. Using the input-output dynamic data recorded during an earthquake event, batch and recursive Bayesian estimation methods are employed to update a mechanics-based nonlinear FE model of the structure of interest (building, bridge, dam, etc.). Unknown parameters of the nonlinear FE model characterizing material constitutive models, inertia, geometric, and/or constraint properties of the structure can be estimated using limited response data recorded through accelerometers or heterogeneous sensor arrays. The updated nonlinear FE model can be used to identify the damage in the structure following a damage-inducing event. This framework, therefore, can provide an advanced tool for post-disaster damage identification and structural health monitoring. The batch estimation method is based on a maximum a posteriori estimation (MAP) approach, where the time history of the input and output measurements are used as a single batch of data for estimating the FE model parameters. This method results in a nonlinear optimization problem that can be solved using gradient-based and non-gradient-based optimization algorithms. In contrast, the recursive Bayesian estimation method processes the information from the measured data recursively, and updates the estimation of the FE model parameters progressively over the time history of the event. The recursive Bayesian estimation method results in a nonlinear Kalman filtering approach. The Extended Kalman filter (EKF) and Unscented Kalman filter (UKF) are employed as recursive Bayesian estimation methods herein. For those estimation methods that require the computation of structural FE response sensitivities (total partial derivatives) with respect to the unknown FE model parameters, the direct differentiation method (DDM) is used. Response data numerically simulated from a nonlinear FE model (with unknown material model parameters) of a five-story two-by-one bay reinforced concrete frame building subjected to bi-directional horizontal seismic excitation are used to illustrate the performance of the proposed framework.