Timothy P. Kernicky
University of North Carolina at Charlotte
Network
Latest external collaboration on country level. Dive into details by clicking on the dots.
Publication
Featured researches published by Timothy P. Kernicky.
Journal of Structural Engineering-asce | 2015
Timothy P. Kernicky; Matthew J. Whelan; David C. Weggel; Corey D. Rice
AbstractStructural identification continues to develop an expanding role within performance-based civil engineering by offering a means to construct high-fidelity analytical models of in-service structures calibrated to experimental field measurements. Although continued advances and case studies are needed to foster the transition of this technique from exploration to practice, potential applications are diverse and range from design validation, construction quality control, assessment of retrofit effectiveness, damage detection, and lifecycle assessment for long-term performance evaluation and structural health monitoring systems. Existing case studies have been primarily focused on large civil structures, specifically bridges, large buildings, and towers, and the limited studies exploring application to damaged structures have been primarily associated with seismic events or other conventional hazards. The current paper produces the first experimental application of structural identification to a compo...
Archive | 2014
Timothy P. Kernicky; Matthew J. Whelan; David C. Weggel
Prestressed concrete has received increased attention as a structural system for blast resistance and protection. However, prestressed concrete members not intentionally designed to resist blast loads, as well as connections to such members, may be especially susceptible to blast initiated damage. Full-scale testing of a prestressed double-tee joist roof in an industrial building was conducted both prior to and after internal detonation of a modestly sized explosive charge located approximately six feet below the base elevation of the double-tee joists. The testing included experimental modal analysis of the roof using a pair of long stroke electrodynamic shakers and a distributed network of 60 accelerometers. A large set of modal parameter estimates are extracted from the measurement data using a combined stochastic-deterministic subspace identification algorithm. Comparisons are made to a numerical model of the roof, developed using properties obtained from supplemental nondestructive evaluation and local historic design handbooks, to assess the plausibility of the modes. Differences in the natural frequencies and mode shapes are highlighted to qualitatively draw conclusions on plausible damage to the roof system alongside physical observations from the site.
Archive | 2016
Timothy P. Kernicky; Matthew J. Whelan
Approaches for structural health monitoring of civil infrastructure using vibration-based damage detection methods have progressed significantly over the past decade as a result of extensive experimentation on both full-scale and laboratory-scale structures. However, the field instrumentation of in-service structures involving well-characterized damage scenarios is a complex, costly, and high-risk investment. In contrast, the use of numerical simulations to generate synthetic data for structural health monitoring research is restricted by challenges associated with modeling experimental uncertainties associated with measurement devices and ambient disturbances. In both cases, the faithful representation of meaningful damage progression is often a technical limitation. Recently, hybrid simulation has been explored as a novel alternative for experimental structural health monitoring research that alleviates the expense and logistics associated with full structural testing, yet retains the empirical nature of experimentation by combining an experimental substructure with an analytical model. In this paper, the use of pseudo-dynamic hybrid simulation for exploration of vibration-based structural health monitoring techniques is demonstrated using a truss structure where a single member and bolted gusset plate connection are incorporated in the experimental model. Hybrid simulations are performed in the baseline state and following introduction of localized damage developed at a limit state of the bolted connection. The results correlate strongly with predictions obtained from a fully analytical model and support the further exploration of hybrid simulation as a tool for vibration-based structural health monitoring research.
Archive | 2014
Matthew J. Whelan; Timothy P. Kernicky; David C. Weggel
Structural identification has continued to develop into a versatile tool for developing high fidelity analytical models of large civil structures that accurately reflect the measured in-service response. The results of successful structural identification have been applied to validate the performance of innovative systems and improve assessments of response analysis for operational and extreme loads. Furthermore, the developing field of vibration-based damage detection has sought to employ structural identification for long-term performance monitoring and condition assessment of aged structures. Overwhelmingly, the finite element method has served as the analytical framework for such models. However, alternative physics engines, such as the Applied Element Method, offer distinct advantages over the finite element method both with respect to the computational considerations in the identification process and with respect to the use of the calibrated model for assessment of structural response to extreme loads. A general framework for structural identification with applied elements is discussed, and advantages are contrasted with traditional finite element approaches. A case study application, a prestressed concrete double-tee joist roof tested in a full-scale building, is presented to demonstrate the approach and emphasize these advantages.
Structural Health Monitoring-an International Journal | 2018
Timothy P. Kernicky; Matthew J. Whelan; Ehab Al-Shaer
Structural identification has received increased attention over recent years for performance-based structural assessment and health monitoring. Recently, an approach for formulating the finite element model updating problem as a constraint satisfaction problem has been developed. In contrast to widely used probabilistic model updating through Bayesian inference methods, the technique naturally accounts for measurement and modeling errors through the use of interval arithmetic to determine the set of all feasible solutions to the partially described and incompletely measured inverse eigenvalue problem. This article presents extensions of the constraint satisfaction approach permitting the application to larger multiple degree-of-freedom system models. To accommodate for the drastic increase in the dimensionality of the inverse problem, the extended methodology replaces computation of the complete set of solutions with an approach that contracts the initial search space to the interval hull, which encompasses the complete set of feasible solutions with a single interval vector solution. The capabilities are demonstrated using vibration data acquired through hybrid simulation of a 45-degree-of-freedom planar truss, where a two-bar specimen with bolted connections representing a single member of the truss serves as the experimental substructure. Structural identification is performed using data acquired with the undamaged experimental member as well as over a number of damage scenarios with progressively increased severity developed by exceeding a limit-state capacity of the member. Interval hull solutions obtained through application of the nonlinear constraint satisfaction methodology demonstrate the capability to correctly identify and quantify the extent of the damage in the truss while incorporating measurement uncertainties in the parameter identification.
Structural Health Monitoring-an International Journal | 2015
Timothy P. Kernicky; Matthew J. Whelan; Usman Rauf; Ehab Al-Shaer
Physics-based approaches to vibration-based structural health monitoring largely rely on structural identification, or model updating, of a finite element model through correlation with the experimentally measured natural frequencies and mode shapes. Currently, the predominant technique used to perform model updating relies on local optimization strategies, such as gradient-based methods, or global optimization techniques, such as genetic algorithms. However, optimization-based approaches provide limited capabilities for addressing fundamental issues related to the inverse eigenvalue problem, including solution uniqueness, generation of alternative solutions in the presence of measurement uncertainties, and computational efficiency. Nonlinear constraint satisfaction processors with interval arithmetic have been recently explored by the authors as an alternative to optimization techniques for structural identification and provide unique and computationally swift capabilities for addressing these challenges. However, damage detection and diagnostics using structural identification by nonlinear constraint satisfaction processing with experimental data has yet to be explored. In this paper, a series of progressive damages are applied to an instrumented laboratory model of a multi-story shear building. Strategies for damage detection and quantification using finite element model updating with modal parameter estimates and reduced order analytical models are presented alongside results from application to the laboratory data. doi: 10.12783/SHM2015/175
Archive | 2015
Usman Rauf; Timothy P. Kernicky; Matthew J. Whelan; Ehab Al-Shaer
Structural identification of civil infrastructures, using measured modal properties, remains a promising research field with many applications in performance-based civil engineering and structural health monitoring. In particular, either computationally swift or direct methods for identifying structural models from partially described and incomplete modal parameter estimates are of foremost interest to facilitate near real-time and reliable structural performance assessment and diagnostics. This paper proposes modeling structural systems as Constraint Satisfaction Problems (CSPs) for structural identification to solve for uncertain parameters in structural models. Consistent with measurement data, modal parameter estimates are treated as truncated both in terms of the number of modes measured and the number of measured degrees of freedom relative to the analytical model, which yields a challenging nonlinear inverse eigenvalue problem. Using nonlinear constraints and parameter bounds, the Constraint Programming approach is demonstrated to be capable of properly reconstructing estimates of both uncertain structural parameters and unmeasured modal parameters for a truss model with only a limited number of measured degrees of freedom.
Archive | 2015
Timothy P. Kernicky; Matthew J. Whelan; Cristopher D. Moen
In-service condition assessment of large civil infrastructure systems has remained a particularly challenging area of research in the fields of nondestructive evaluation and structural health monitoring (SHM). Extensions of theoretically-based and laboratory verified vibration-based methods for assessing damage have been investigated experimentally on full-scale structures within several studies offering mixed conclusions. This paper introduces a recent experimental test program conducted on a full-scale bridge beam subjected to prescribed damage to the tension reinforcement. Details of the experimental testing program and vibration testing of the full-scale bridge beam both prior to and after damage to tension reinforcement are presented. System identification is applied to compare estimates of the natural frequencies, relative damping factors, and mode shapes obtained in the as-built state against those obtained after cutting over half of the tension reinforcement strands in the beam. A data-driven damage detection algorithm previously applied to detect damage in a full-scale bridge is also explored for application to the current dataset.
Proceedings of SPIE | 2012
Timothy P. Kernicky; Matthew J. Whelan
Structural health monitoring methodologies devised over the past two decades have increasing shown improved robustness in capability to identify the onset of structural damage and locate the source of the damage. However, the pathway to prognostication and life-cycle assessment through structural health monitoring remains stalled by a lack of success in the diagnostic step of experimentally quantifying the severity of damage in suitable, engineering quantities. Of the methods devised, strain energy approaches have demonstrated not only strength in identifying and localizing structural damage but also uniquely provide a theoretical basis for quantifying damage through measurement of relative stiffness loss in individual members. Conventional applications of strain energy methods use distributed accelerometers, often being single-axis and oriented in the same direction. The limited degrees-of-freedom measured limits the modal parameter extraction to a reduced subset and yields only partial reconstruction of the strain energy in the system. Furthermore, it has been shown experimentally and proven analytically that improvement in strain energy methods through increased spatial density of the sampling array is constrained by the effect of measurement noise on the accuracy of the numerical computations. In this paper, alternative sensor topologies are explored for improving the reconstruction of strain energy estimates. An experimental component of the research includes strain energy estimates for a fixed-free beam heavily instrumented with accelerometers. Prescribed damage is incrementally applied to the beam to permit a basis for comparison amongst the sensor topologies in addressing the damage diagnostics problem with specific emphasis on quantification of severity through stiffness loss.
Computers & Structures | 2017
Timothy P. Kernicky; Matthew J. Whelan; Usman Rauf; Ehab Al-Shaer