Georgia Institute of Technology
Data-Driven and GPU-Accelerated Computational Methods for High-Fidelity Aerostructural Design
Abstract
dc:description.abstractHigh-fidelity aerostructural optimization is critical for advanced aerospace design. One key bottleneck for highly flexible wings is the cost of geometric nonlinear structural analysis. This work compares 14 state-of-the-art GPU-accelerated linear solvers for nonlinear structural analysis with Reissner-Mindlin shell elements. Domain decomposition methods are found to be more robust across thick and thin shell designs than other multilevel solvers. A novel wraparound subdomain splitting method leads to stabilized thin Mindlin shell performance on wings, leading to 8 × −100× speedup compared to other subdomain splittings. The co-design of Reissner-Mindlin shell elements and multilevel solvers are also explored for thickness-independent or Kirchoff-like performance. A new mixed isogeometric element is given for cylindrical shells with 5× speedup compared to with standard element types. Machine learning methods are also explored to improve the fidelity of nonlinear constraints in aerostructural design, including applications to stiffened panel buckling and time-series classification of propulsion system anomalies.
Degree
thesis:*- Name thesis:degree_name
- Aerospace Engineering, PhD
- Department thesis:degree_department
- Aerospace Engineering
- Grantor
- Georgia Institute of Technology
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Engelstad, Sean Phillip
- Advisor dc:contributor.advisor
-
- Kennedy, Graeme J.
- Committee members dc:contributor.committeemember
-
- Qian, Elizabeth
- James, Kai
- Athanasiou, Christos
- ORCID person:identifier.orcid
Rights
- Language dc:language.iso
- English
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/81753