Clemson University
A Novel Computationally Efficient AI-Driven Generative Inverse Design Framework for Accelerating Topology Optimization and Designing Lattice-Infused Structures
Abstract
dc:description.abstractMultiscale topology optimization (TO) provides an inverse design computational framework for designing globally and locally optimized hierarchical structures. Triply periodic minimal surfaces (TPMS), a subclass of parametrically-driven lattice structures, exhibit unique properties such as large surface area, significant volume densities, and good strength-to-weight ratio, which makes them favorable for novel engineering applications. The recent advances in additive manufacturing and its ability to fabricate high-resolution structures have spurred interest in multiscale TO and TPMS for computationally designing finer and high-resolution designs. While multiscale TO and TPMS bring transformative opportunities in various applications, their potential for everyday use remains idle due to critical issues associated with their frameworks. The multiscale TO computational paradigm brings new challenges, including geometric frustration, non-smooth boundaries, and higher computational time that needs to be handled. Further developing design tools for TPMS brings in additional challenges such as limited design space and lack of inverse design tools for targeted properties to geometries mapping. This dissertation attempts to address these above-mentioned challenges by: (1) Employing a computationally efficient deep learning-based TO pipeline for microscale TO analysis. (2) Developing a novel CNN-based connectivity improvement solver for improving micro-structural connectivity between neighboring unit cells. (3) Utilizing a post-processing solver for neural network outputted structures to ensure physical constraints, optimized material distribution, and proper local connectivity. (4) Proposing a weighted combinatorial approach for developing novel TPMS-based lattices from baseline TPMS structures such as Schwarz-P, Diamond-D, and Schoen's F-RD surfaces. (5) Developing deep learning-based inverse design framework for predicting TPMS-based lattices for targeted effective elastic properties. (6) Developing deep learning-based multiscale TO framework for TPMS lattices-infused structures.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy (PhD)
- Level thesis:degree_level
- Dissertation
- Discipline thesis:degree_discipline
- Automotive Engineering
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Patel, Darshil
- Contributors dc:contributor
-
- Dr. Rahul Rai
- Dr. Gary Dargush
- Dr. Venkat Krovi
- Dr. Srikant Pilla
Subjects
dc:subject × 7Identifiers
dc:identifier.*- Repository record dc:identifier
- https://open.clemson.edu/all_dissertations/3138