Announcement Detail

Wednesday, December 3, 2025

1:00 PM CST

Join via Zoom: https://us06web.zoom.us/j/87331120699?pwd=MExkQ085ZHh1aDc3ZnBwQzFlUHI4UT09

Large Scale Structural Systems and Optimal Design Industrial Colloquium

Leveraging polygonal primitives in feature-mapping topology optimization for manufacturing constraints

Speaker

Yakov Zelickman, Johns Hopkins University

Abstract

Feature mapping approaches in topology optimization offer a compelling compromise between the simplicity of density-based formulations and the geometric control of shape optimization. However, existing implementations typically rely on primitives with analytical signed distance functions (SDF), such as circles or super-ellipses, limiting the geometrical expressiveness and the range of manufacturability constraints that can be directly imposed.

In this talk, I will present a general polygonal primitive (PP) formulation that extends feature mapping to arbitrary convex and non-convex shapes with any number of vertices. Each vertex coordinate is treated as a design variable, providing a fully explicit geometric parametrization. A new differentiable and computationally efficient SDF approximation is introduced for arbitrary polygons, accompanied by a self- intersection prevention scheme that guarantees geometric validity throughout the optimization.

This parametrization allows direct integration of manufacturing constraints, including minimum and maximum feature sizes, overhang restrictions, and alignment or size limits, which might be more challenging or with other formulations such density-based TO. Numerical results demonstrate that optimized designs using polygonal primitives achieve performance levels comparable to classical density-based methods while offering improved geometric clarity and manufacturability. The inherent straight-edged boundaries are particularly advantageous in fabrication processes such as reinforced concrete casting and machining, where curved geometries are impractical.

Overall, the proposed polygonal primitive framework provides a geometrically rich, differentiable, and manufacturable extension to existing feature-mapping methods, bridging the gap between high-resolution topology optimization and practical structural design.

Bio

Yakov Zelickman is an Assistant Research Professor in the Department of Civil and Systems Engineering. He has a theoretical background in solid and fracture mechanics, non-linear structural analysis, and optimization, and extensive practical experience as a structural engineer. His research focuses on developing structural optimization algorithms aimed at creating smarter and more efficient structures.

Zelickman is currently working on several projects in the field of structural design and optimization, which span different system scales and a variety of physical phenomena. His work includes leveraging optimization techniques to explore new architectures for structural batteries, material discovery and characterization, and concrete structural systems.

Zelickman has been honored with the Nezer Prize for outstanding research in the analysis and design of structures. He is also an active member of the Optimal Structural Design Committee of the American Society of Civil Engineer’s Structural Engineering Institute, underscoring his dedication to development of the field of structural optimization and outreach in the engineering community.

Prior to joining Johns Hopkins faculty, Zelickman completed his postdoctoral fellowship with JHU’s Department of Civil and Systems Engineering. He earned his BSc, MSc, and PhD—all in structural engineering—from the Technion – Israel Institute of Technology.