USACM Large Scale Structural Systems and Optimal Design Colloquium - Julian Norato

April 1, 2026 Julian Norato, University of Connecticut Topology optimization with geometric primitives to tackle some large-scale design problems Density-based and level-set techniques for topology optimization are powerful tools for the computational design of structures and architected materials, and they have been used with success across many physical domains. However, their verbose representation of the design –typically requiring thousands to millions of design variables—makes their application to some large-scale problems challenging. By contrast, feature-mapping techniques, like the geometry projection method, represent the structure using geometric primitives and typically require at most tens of design variables. This representation conciseness opens the door to solve some large-scale problems that have a prohibitive cost with field representations. In this talk, I will present several ways in which geometry projection techniques can be used to solve some large-scale topology optimization problems, including: adaptive mesh refinement to design plate structures made of isotropic and fiber-reinforced materials, whereby the plate thickness is much smaller than the dimensions of the design region; design of ultra-light frames (defined as having a volume of no more than 1% of the volume of the design region) by defining primitives equivalent to structural profiles; the use of off-the-shelf and parallel finite-difference sensitivities to design plate heat exchangers for improved heat transfer and sandwich panels for improved energy absorption; and the reliability-based design of structures with regard to variability in the dimensions and position of the structural members.