|Tipo di tesi||Tesi di dottorato di ricerca|
|Titolo||Ottimizzazione multi-obiettivo di modelli agli elementi finiti per l’identificazione di parametri meccanici e geometrici in applicazioni di ingegneria civile|
|Titolo in inglese||Multi-objective optimization of finite element models to identify mechanical and geometrical parameters in civil engineering applications|
|Settore scientifico disciplinare||ICAR/09 - TECNICA DELLE COSTRUZIONI|
|Corso di studi||INGEGNERIA INDUSTRIALE E DEL TERRITORIO|
|Data inizio appello||2019-03-08|
|Disponibilità||Accessibile via web (tutti i file della tesi sono accessibili)|
Molti problemi di ingegneria strutturale vengono affrontati grazie alla realizzazione di modelli agli elementi finiti. I modelli numerici spesso presentano imprecisioni e approssimazioni dovute all’incertezza con cui si conoscono alcuni parametri che possono inficiare il corretto ottenimento dei risultati cercati. Per identificare le variabili incognite, è necessario ottimizzare il modello, massimizzando o minimizzando una funzione obiettivo.
Many structural problems in civil engineering are studied with finite element models. These models are often subjected to imprecisions because several parameters are uncertain, and, therefore, results may be uncorrect. Structural optimization allows to identify unknown variables by means maximizing or minimizing an objective function. The optimization procedure requires the identification of unknown variables that drive the problem and the achievement of many objectives simultaneously. Objectives often conflict each other and multi-objective optimizations have not easy solutions. In many cases, multi-objective problems have not a unique solution, but they have a set of optimal points that represents the Pareto front. The front is the set of solutions that are the best compromise among conflicting objectives. A multi-objective problem may be solved by combining the objectives into a single-objective function by using the so-called weighted sum method. The optimal points of the Pareto front may be identified by changing weighting factors. A posteriori criterion has to be applied to find the preferred solution among the Pareto front solutions. An optimization procedure with a proper optimization algorithm is required to efficiently solve multi-objective problems. The algorithm has to minimize the computational effort, while the optimization procedure automitizes user's operations. In this work, a customized optimization procedure is presented. The procedure allows to transfer data from an optimization algorithm to a finite element software. In particular, a Python script allows the passage of data from the DE-S algorithm, implemented in Matlab environment, and the finite element model realized in Abaqus. The optimization procedure has been applied in two case studies. In the first, the finite element model of an ancient damaged fortress has been optimized on the basis of experimental dynamic test results. Once its dynamic properties were identified, the model has been optimized in order to minimize the objective function that measures the difference between numerical and experimental modal properties. The optimization process has identified some unknown mechanical parameters, as the elastic modulus of cracked masonry. In the second study, the shape of an energy dissipating steel connector in precast walls has been optimized. The shape of the transversal section has been optimized in order to obtain the structural performances required as stiffness, resistance, and energy dissipation. Finally, the optimization results of the transversal section will be verified by performing experimental tests on prototypes of steel connector.