In recent decades, the gradual integration of approaches that rely on mathematical or algorithmic logic into specific phases of the design process—along with the emergence of a cognitive, system-wide perspective associated with the concept of Computational Design—has brought about a methodological revolution in the AEC (Architecture, Engineering, and Construction) sector. This paradigm shift has enabled designers to formulate complex problems and explore solution spaces in a more systematic and innovative manner. This thesis is structured in two parts. The first consists of an investigation aimed at identifying recurrent operational strategies and application domains of Computational Design through the analysis of approximately thirty architectural and engineering projects. The objective is to propose improvements to conventional design workflows by introducing computational methodologies. Many of the strategies examined concern the identification of morphologically “optimal” configurations within a design system, a practice commonly known as form-finding. When addressing complex problems, where multiple heterogeneous factors must be considered simultaneously, single-objective analyses often prove inadequate for capturing the intrinsic complexity of real-world systems. It thus becomes necessary to adopt multi-objective optimization processes capable of balancing divergent or competing performance criteria. The second part of the thesis focuses on the analysis of optimization processes that integrate concurrent simulations in the search for solutions to morphogenetic design problems. The objects of investigation are complex hyperboloid geometries, which—due to their established historical use—represent one of the most virtuous examples of the synthesis between form and function within performance-based design frameworks. Following an in-depth examination of the various generation methods and the geometric-analytical relationships governing the behavior of these shapes, a parametric model was developed for their digital representation. This modeling process took place within Rhinoceros 3D, a CAD environment based on NURBS surface representation, while parametric control was implemented through Grasshopper, a software platform employing visual programming logic. The generated geometries were subjected to both structural analyses—using finite element (FEM) and isogeometric (IGA) approaches—and computational fluid dynamics (CFD) simulations, with the aim of identifying meaningful performance indicators for evaluating the efficiency of the proposed design solutions. The resulting criteria, along with the fundamental parameters governing the parametric model, were used to formulate a multi-objective optimization problem, which was solved using genetic algorithms. The solutions obtained underwent a filtering phase and were subsequently evaluated through both qualitative and quantitative criteria. This led to a critical assessment of the actual validity of the adopted methods, contributing to a broader reflection on the disciplinary foundations and future potential of Computational Design.
Negli ultimi decenni, l’introduzione progressiva nel processo progettuale di approcci che richiedono l’uso della logica matematica o algoritmica nella formalizzazione di alcune specifiche fasi, nonché di una visione cognitiva d’insieme riconducibili al concetto di Computational Design, ha determinato una rivoluzione metodologica nel settore AEC, che ha permesso ai progettisti di formulare problemi complessi ed esplorare lo spazio delle soluzioni in modo innovativo e sistematico. La presente tesi si articola in due fasi. La prima consiste in indagine condotta al fine di individuare possibili strategie operative ricorrenti e ambiti applicativi del Computational Design, mediante l’analisi di circa trenta progetti architettonici e ingegneristici. L’obiettivo è proporre possibili miglioramenti del flusso progettuale attraverso l’introduzione di metodologie computazionali. Numerose tra le strategie esaminate hanno riguardato l’individuazione di configurazioni morfologiche “ottimali” di un sistema progettuale, prassi comunemente nota come form-finding. In problemi natura complessa, la risoluzione richiede la considerazione simultanea di molteplici fattori eterogenei in cui l’impiego di analisi mono-obiettivo, risulta inadeguato nel rappresentare la complessità di sistemi reali. Si rende pertanto necessario ricorrere a processi di ottimizzazione multi-obbiettivo, capaci di gestire prestazioni divergenti o concorrenti. La seconda parte del lavoro è dedicata all’analisi di processi di ottimizzazione che integrano simulazioni concorrenti nella ricerca di soluzioni al problema di definizione morfologica posto dai progettisti. Oggetto di tale processo sono le geometrie iperboloidiche complesse, che, per consolidato impiego storico, rappresentano uno dei più virtuosi esempi di connubio tra forma e funzione all’interno di un quadro simulativo di specifiche prestazioni. Dopo l’approfondimento delle molteplici modalità di generazione e delle relazioni geometrico-analitiche che regolano il comportamento di tali figure, è stato realizzato un modello parametrico per la loro riproduzione digitale. L’operazione è avvenuta all’interno di Rhinoceros3D, un ambiente CAD che basa la rappresentazione delle superfici mediante N.U.R.B.S. mentre la parametrizzazione del modello è stata realizzata attraverso Grasshopper, software che utilizza linguaggio di programmazione visuale. Le geometrie ottenute sono state sottoposte simultaneamente ad analisi strutturali -mediante approcci agli elementi finiti (FEM) o isogeometrici (IGA)- e ad analisi fluidodinamiche, finalizzate all’individuazione di parametri significativi per la valutazione dell’efficienza delle soluzioni generate. I criteri risultanti, insieme ai parametri fondamentali che regolano il funzionamento del modello parametrico, sono utilizzati nella formulazione di un problema di ottimizzazione multi-obiettivo, la cui risoluzione è avvenuta mediante l’utilizzo di algoritmi genetici. Le soluzioni, oggetto di una fase di filtraggio, hanno infine condotto a considerazioni qualitative e quantitative riguardanti l’effettiva validità dei metodi adottati in un’ottica di giudizio generale sull’intera disciplina del Computational Design.
Analisi dei processi di ottimizzazione tramite computational design. Il caso di simulazioni concorrenti in geometrie complesse
DE BIASIO, GIACOMO
2024/2025
Abstract
In recent decades, the gradual integration of approaches that rely on mathematical or algorithmic logic into specific phases of the design process—along with the emergence of a cognitive, system-wide perspective associated with the concept of Computational Design—has brought about a methodological revolution in the AEC (Architecture, Engineering, and Construction) sector. This paradigm shift has enabled designers to formulate complex problems and explore solution spaces in a more systematic and innovative manner. This thesis is structured in two parts. The first consists of an investigation aimed at identifying recurrent operational strategies and application domains of Computational Design through the analysis of approximately thirty architectural and engineering projects. The objective is to propose improvements to conventional design workflows by introducing computational methodologies. Many of the strategies examined concern the identification of morphologically “optimal” configurations within a design system, a practice commonly known as form-finding. When addressing complex problems, where multiple heterogeneous factors must be considered simultaneously, single-objective analyses often prove inadequate for capturing the intrinsic complexity of real-world systems. It thus becomes necessary to adopt multi-objective optimization processes capable of balancing divergent or competing performance criteria. The second part of the thesis focuses on the analysis of optimization processes that integrate concurrent simulations in the search for solutions to morphogenetic design problems. The objects of investigation are complex hyperboloid geometries, which—due to their established historical use—represent one of the most virtuous examples of the synthesis between form and function within performance-based design frameworks. Following an in-depth examination of the various generation methods and the geometric-analytical relationships governing the behavior of these shapes, a parametric model was developed for their digital representation. This modeling process took place within Rhinoceros 3D, a CAD environment based on NURBS surface representation, while parametric control was implemented through Grasshopper, a software platform employing visual programming logic. The generated geometries were subjected to both structural analyses—using finite element (FEM) and isogeometric (IGA) approaches—and computational fluid dynamics (CFD) simulations, with the aim of identifying meaningful performance indicators for evaluating the efficiency of the proposed design solutions. The resulting criteria, along with the fundamental parameters governing the parametric model, were used to formulate a multi-objective optimization problem, which was solved using genetic algorithms. The solutions obtained underwent a filtering phase and were subsequently evaluated through both qualitative and quantitative criteria. This led to a critical assessment of the actual validity of the adopted methods, contributing to a broader reflection on the disciplinary foundations and future potential of Computational Design.| File | Dimensione | Formato | |
|---|---|---|---|
|
DeBiasio_Analisi dei processi di ottimizzazione.pdf
embargo fino al 15/01/2027
Dimensione
37.07 MB
Formato
Adobe PDF
|
37.07 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/89157