Firus, Andrei (2022)
A contribution to moving force identification in bridge dynamics.
Technische Universität Darmstadt
doi: 10.26083/tuprints-00020293
Dissertation, Erstveröffentlichung, Verlagsversion
Kurzbeschreibung (Abstract)
In the actual context of an aging infrastructure and a clear trend towards ever faster and heavier vehicles, the reassessment of existing bridges becomes an ever more challenging task for civil engineers, especially when crucial decisions regarding costly replacement or rehabilitation measures have to be made. An accurate knowledge of the real forces acting on the structure offers the engineers the possibility of a less conservative, but still reliable reassessment process, which can lead to important extensions of the remaining service life and thus to considerable financial benefits for both the bridge owners and the society. However, the direct measurement of dynamic forces acting on bridges is in most cases either not possible or very cumbersome. On the contrary, time histories of the structural responses can be recorded with reasonable effort.
The present work presents a novel approach for identification of dynamic moving forces acting on a bridge structure. It seeks to find the optimal force values that minimize the difference between the computed and measured displacement and acceleration time histories for a limited number of sensor locations. For this, an optimization problem is formulated in time domain. It considers the magnitude of each moving force in each time instant as an unknown variable. The regularization of the inverse problem is addressed by introducing two penalty functions that define physical a priori information on the solution. The problem is solved by means of the sensitivity-based trust region optimization algorithm, which requires the exact gradient and Hessian matrix of the objective function. These are determined by analytical differentiation, whereas the time stepping method of Newmark is used to solve a set of differential equations occurring in the implicit differentiation process. Moreover, an efficient solution of problems with a large number of unknown force values is possible by using a sliding window approach. The proposed moving identification method is applicable to any 3D Finite Element model with arbitrary level of detail. This is achieved by employing the principle of modal superposition.
A comprehensive study with simulated measurement data was firstly conducted to analyze the isolated effects of various parameters on the identification accuracy of the proposed method. In addition, an experimental validation was performed based on two investigations: an existing railway bridge in operation subjected to the passages of an ICE 4 vehicle and an experimental full-scale pedestrian bridge. They underline the wide range of possible applications for the moving force identification method developed within this work: from small structures excited by rather low pedestrian forces up to the "heavy category" of a complete train passing a railway bridge in operation. The results are very satisfactory, consistently indicating that the proposed method allows a sufficiently accurate identification of the moving forces. In this context, a set of particularities and limitations that arise in the practical the application of the method on real structures are also discussed. These are mostly related to modeling inaccuracies, which have been identified as a crucial factor influencing the solution behavior.
Typ des Eintrags: | Dissertation | ||||
---|---|---|---|---|---|
Erschienen: | 2022 | ||||
Autor(en): | Firus, Andrei | ||||
Art des Eintrags: | Erstveröffentlichung | ||||
Titel: | A contribution to moving force identification in bridge dynamics | ||||
Sprache: | Englisch | ||||
Referenten: | Schneider, Prof. Dr. Jens ; Lombaert, Prof. Dr. Geert ; Kemmler, Prof. Dr. Roman | ||||
Publikationsjahr: | 2022 | ||||
Ort: | Darmstadt | ||||
Kollation: | xxx, 335 Seiten | ||||
Datum der mündlichen Prüfung: | 17 Dezember 2021 | ||||
DOI: | 10.26083/tuprints-00020293 | ||||
URL / URN: | https://tuprints.ulb.tu-darmstadt.de/20293 | ||||
Kurzbeschreibung (Abstract): | In the actual context of an aging infrastructure and a clear trend towards ever faster and heavier vehicles, the reassessment of existing bridges becomes an ever more challenging task for civil engineers, especially when crucial decisions regarding costly replacement or rehabilitation measures have to be made. An accurate knowledge of the real forces acting on the structure offers the engineers the possibility of a less conservative, but still reliable reassessment process, which can lead to important extensions of the remaining service life and thus to considerable financial benefits for both the bridge owners and the society. However, the direct measurement of dynamic forces acting on bridges is in most cases either not possible or very cumbersome. On the contrary, time histories of the structural responses can be recorded with reasonable effort. The present work presents a novel approach for identification of dynamic moving forces acting on a bridge structure. It seeks to find the optimal force values that minimize the difference between the computed and measured displacement and acceleration time histories for a limited number of sensor locations. For this, an optimization problem is formulated in time domain. It considers the magnitude of each moving force in each time instant as an unknown variable. The regularization of the inverse problem is addressed by introducing two penalty functions that define physical a priori information on the solution. The problem is solved by means of the sensitivity-based trust region optimization algorithm, which requires the exact gradient and Hessian matrix of the objective function. These are determined by analytical differentiation, whereas the time stepping method of Newmark is used to solve a set of differential equations occurring in the implicit differentiation process. Moreover, an efficient solution of problems with a large number of unknown force values is possible by using a sliding window approach. The proposed moving identification method is applicable to any 3D Finite Element model with arbitrary level of detail. This is achieved by employing the principle of modal superposition. A comprehensive study with simulated measurement data was firstly conducted to analyze the isolated effects of various parameters on the identification accuracy of the proposed method. In addition, an experimental validation was performed based on two investigations: an existing railway bridge in operation subjected to the passages of an ICE 4 vehicle and an experimental full-scale pedestrian bridge. They underline the wide range of possible applications for the moving force identification method developed within this work: from small structures excited by rather low pedestrian forces up to the "heavy category" of a complete train passing a railway bridge in operation. The results are very satisfactory, consistently indicating that the proposed method allows a sufficiently accurate identification of the moving forces. In this context, a set of particularities and limitations that arise in the practical the application of the method on real structures are also discussed. These are mostly related to modeling inaccuracies, which have been identified as a crucial factor influencing the solution behavior. |
||||
Alternatives oder übersetztes Abstract: |
|
||||
Status: | Verlagsversion | ||||
URN: | urn:nbn:de:tuda-tuprints-202932 | ||||
Sachgruppe der Dewey Dezimalklassifikatin (DDC): | 600 Technik, Medizin, angewandte Wissenschaften > 620 Ingenieurwissenschaften und Maschinenbau | ||||
Fachbereich(e)/-gebiet(e): | 13 Fachbereich Bau- und Umweltingenieurwissenschaften 13 Fachbereich Bau- und Umweltingenieurwissenschaften > Institut für Statik und Konstruktion 13 Fachbereich Bau- und Umweltingenieurwissenschaften > Institut für Statik und Konstruktion > Fachgebiet Statik |
||||
Hinterlegungsdatum: | 09 Feb 2022 15:16 | ||||
Letzte Änderung: | 10 Feb 2022 07:34 | ||||
PPN: | |||||
Referenten: | Schneider, Prof. Dr. Jens ; Lombaert, Prof. Dr. Geert ; Kemmler, Prof. Dr. Roman | ||||
Datum der mündlichen Prüfung / Verteidigung / mdl. Prüfung: | 17 Dezember 2021 | ||||
Export: | |||||
Suche nach Titel in: | TUfind oder in Google |
Frage zum Eintrag |
Optionen (nur für Redakteure)
Redaktionelle Details anzeigen |