Loading...
Multiscale topology optimization of composite materials for additive manufacturing
Islam, Md Mohaiminul
Islam, Md Mohaiminul
Citations
Altmetric:
Genre
Thesis/Dissertation
Date
2025-12
Advisor
Committee member
Group
Department
Mechanical Engineering
Permanent link to this record
Collections
Files
Research Projects
Organizational Units
Journal Issue
DOI
https://doi.org/10.34944/1p2m-3b38
Abstract
In an era where sustainable and high-performance engineering solutions are critically needed, this dissertation introduces a transformative framework that redefines structural design through the synergistic integration of topology optimization (TO), composite additive manufacturing (AM), and artificial intelligence (AI). This research responds to the urgent demands of the aerospace, automotive, biomedical, renewable energy, and infrastructure industries for lightweight, durable, and eco-efficient structural components. By merging advanced computational design with cutting-edge manufacturing technologies, the framework establishes a new standard for optimized performance, manufacturability, and environmental sustainability—laying a compelling foundation for innovation in modern engineering.
At the core of this work is a novel topology optimization methodology tailored for fiber reinforced polymer composites (FRPCs), materials increasingly used in high-performance applications due to their superior strength-to-weight ratios and customizable mechanical properties. Unlike conventional TO methods that often neglect material anisotropy and fabrication constraints, the proposed approach is specifically designed for advanced composite AM processes. Techniques such as continuous fiber-reinforced 3D printing and 3D Fiber Tethering (3DFIT) are leveraged to fabricate structurally optimized components while addressing critical manufacturing limitations, including discrete fiber orientations, scaffold support structures, and build sequence constraints. A key innovation is the introduction of a Penalized Normal Distribution (PND) scheme that enables simultaneous optimization of structural topology and fiber paths. This technique drives continuous orientation variables toward discrete, manufacturing-compatible directions with high accuracy and computational efficiency—achieving superior convergence speed, fiber orientation fidelity, and structural compliance in both 2D and 3D design problems.
A representative demonstration of the framework is presented through the crashworthy design of an automotive B-pillar, a component subject to strict performance and weight-reduction requirements. The co-optimization of geometry and fiber placement yields significant material savings without compromising mechanical integrity. The optimized B-pillar exhibits enhanced crash energy absorption and improved load distribution, highlighting the real-world viability of this method. Moreover, the reduced material usage and precise structural tailoring support sustainability goals by minimizing waste and enabling greener fabrication pathways.
To further enhance the framework’s scalability and address the high computational cost inherent in traditional TO workflows, an AI-driven acceleration strategy is introduced. At its core is a hybrid loss function that integrates geometric similarity measures with distance-based metrics, guiding neural networks toward accurate topology predictions. This hybrid loss improves structural connectivity, reduces compliance error, and ensures manufacturable outputs without post-processing.
Building on this, a dual-stage ResUNet architecture with conditional physics coupling is proposed to refine design predictions. In this approach, the first ResUNet model produces an initial low-resolution topology estimate, while the second network corrects structural disconnections and enforces physical constraints through an embedded conditioning mechanism. This strategy preserves topological validity and mechanical performance while substantially reducing iteration times. The combined AI framework accelerates topology optimization by up to 400× compared to traditional solvers, enabling real-time design iteration and decision-making across multidisciplinary applications.
Together, these contributions establish an integrated, intelligent, and manufacturing-aware structural optimization pipeline. By embedding physical principles into data-driven algorithms and customizing optimization strategies to accommodate advanced manufacturing constraints, this work
advances not only the science of structural design but also the practice of sustainable engineering. It envisions a future in which design is no longer limited by trade-offs between performance, manufacturability, and computational cost, but instead driven by holistic, adaptive systems that combine the strengths of AI, material innovation, and computational mechanics. The result is a scalable pathway toward resilient, efficient, and high-performance composite structures for the challenges of the 21st century.
Description
Citation
Citation to related work
Has part
ADA compliance
For Americans with Disabilities Act (ADA) accommodation, including help with reading this content, please contact scholarshare@temple.edu
Embedded videos
License
IN COPYRIGHT- This Rights Statement can be used for an Item that is in copyright. Using this statement implies that the organization making this Item available has determined that the Item is in copyright and either is the rights-holder, has obtained permission from the rights-holder(s) to make their Work(s) available, or makes the Item available under an exception or limitation to copyright (including Fair Use) that entitles it to make the Item available.