This dissertation presents an innovative framework that integrates artificial intelligence (AI), specifically automation and machine learning (ML), into the architectural and structural early design of tall buildings with outer diagrids, focusing on seismic resilience. At the core of this research is developing an AI-enhanced methodology that encapsulates computational design, structural engineering, and advanced ML tools, including creating and validating surrogate models for efficient seismic response prediction. A key advancement is the automated workflow that synergizes various software tools—Rhinoceros 3D for architectural modeling, Grasshopper for parametric design, Karamba 3D and OpenSees for structural analysis, and Python scripts for simulation automation—culminating in a seamless, integrated design process. This workflow is further enhanced by leveraging high-performance computing, significantly accelerating the simulation of seismic responses across multiple design iterations. This research also investigates the interplay between architectural parameters (input features) and structural performance metrics (responses to seismic events) through techniques like feature selection (to find the most important parameters). This work emphasizes surrogate modeling, utilizing both classical ML and deep learning techniques, to overcome the computational burden of traditional finite element analysis, offering substantial efficiency without sacrificing accuracy. Surrogate models are trained based on input features and responses, providing a deeper understanding of how design decisions impact seismic resilience. Through meticulous case studies, the framework's practicality is demonstrated, showcasing its ability to harmonize architectural aesthetics with structural integrity under seismic forces. This dissertation contributes an innovative approach to tall building design, underscoring the transformative potential of AI in facilitating informed, resilient, and efficient architectural and structural solutions from the conceptual stage. By highlighting the integration of AI through automation and ML-based surrogate modeling, this work paves the way for future advancements in the field.
Questa tesi presenta un approccio innovativo che integra l'intelligenza artificiale (IA) nella progettazione preliminare architettonica e strutturale degli edifici alti, concentrandosi sul loro comportamento in presenza di sisma moderato. Al centro di questa ricerca c'è lo sviluppo di una metodologia potenziata dall'IA che include la progettazione computazionale, l'ingegneria strutturale e strumenti avanzati di apprendimento automatico, compresa la creazione e la validazione di modelli surrogati per una previsione efficiente della risposta sismica. Un progresso chiave è il flusso di lavoro automatizzato che comprende in modo sinergico vari strumenti software, Rhinoceros 3D™ per la modellazione architettonica, Grasshopper™ per la progettazione parametrica, Karamba 3D e OpenSees per l'analisi strutturale, e script Python per l'automazione della simulazione, culminando in un processo di progettazione integrato e senza soluzione di continuità. Questo flusso di lavoro è ulteriormente potenziato sfruttando il calcolo ad alte prestazioni, accelerando significativamente la simulazione delle risposte sismiche attraverso molteplici iterazioni di progettazione. La ricerca enfatizza la modellazione surrogata, utilizzando sia tecniche di apprendimento automatico classico che di apprendimento profondo, per superare l'onere computazionale dell'analisi degli elementi finiti tradizionale senza sacrificare l'accuratezza. Attraverso due casi studio, viene dimostrata la praticità del quadro, mostrando la sua capacità di armonizzare l'estetica architettonica con l'integrità strutturale rispetto alle forze sismiche. Questa tesi contribuisce alla progettazione di edifici alti, sottolineando il potenziale trasformativo dell'IA nel produrre soluzioni architettoniche e strutturali resilienti ed efficienti fin dalla fase concettuale. Mettendo in evidenza l'integrazione dell'IA, in particolare attraverso la modellazione surrogata e i flussi di lavoro automatizzati, questo lavoro apre la strada a futuri progressi nel campo.
A workflow for architectural and structural early design of tall buildings with outer diagrids through AI-driven and computational design strategies
Kazemisangedehi, Seyedpooyan
2023/2024
Abstract
This dissertation presents an innovative framework that integrates artificial intelligence (AI), specifically automation and machine learning (ML), into the architectural and structural early design of tall buildings with outer diagrids, focusing on seismic resilience. At the core of this research is developing an AI-enhanced methodology that encapsulates computational design, structural engineering, and advanced ML tools, including creating and validating surrogate models for efficient seismic response prediction. A key advancement is the automated workflow that synergizes various software tools—Rhinoceros 3D for architectural modeling, Grasshopper for parametric design, Karamba 3D and OpenSees for structural analysis, and Python scripts for simulation automation—culminating in a seamless, integrated design process. This workflow is further enhanced by leveraging high-performance computing, significantly accelerating the simulation of seismic responses across multiple design iterations. This research also investigates the interplay between architectural parameters (input features) and structural performance metrics (responses to seismic events) through techniques like feature selection (to find the most important parameters). This work emphasizes surrogate modeling, utilizing both classical ML and deep learning techniques, to overcome the computational burden of traditional finite element analysis, offering substantial efficiency without sacrificing accuracy. Surrogate models are trained based on input features and responses, providing a deeper understanding of how design decisions impact seismic resilience. Through meticulous case studies, the framework's practicality is demonstrated, showcasing its ability to harmonize architectural aesthetics with structural integrity under seismic forces. This dissertation contributes an innovative approach to tall building design, underscoring the transformative potential of AI in facilitating informed, resilient, and efficient architectural and structural solutions from the conceptual stage. By highlighting the integration of AI through automation and ML-based surrogate modeling, this work paves the way for future advancements in the field.File | Dimensione | Formato | |
---|---|---|---|
Pooyan_Thesis__final_ (submitted_edition).pdf
solo utenti autorizzati a partire dal 24/09/2025
Descrizione: PhD thesis
Dimensione
146.99 MB
Formato
Adobe PDF
|
146.99 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/228432