Dow Chemical | Predictive Analysis and Simulation-Driven Design for Structural Sealant Glazing under Seismic Loading

We were honored to have Hang Shu from Dow Chemical present at the 2026 SIMULIA Americas Users Conference in Novi, Michigan, May 13-14, 2026.

Abstract: 

This work presents an integrated experimental–computational framework to evaluate the seismic performance of Structural Sealant Glazing (SSG) systems and to advance predictive capabilities at both material and assembly levels. High fidelity constitutive models were developed for DOWSIL™ silicone sealants, capturing hyperelasticity, viscoelasticity, and Mullins softening behaviors. Model development combined uniaxial tensile and cyclic loading experiments on coupon specimens with parameter calibration using Abaqus/CAE, Isight, and physics informed neural networks (PINNs). Model validation was achieved via comparison with cyclic lap shear experiments, demonstrating strong agreement in reproducing stress concentrations, strain localization, and damage related softening under repeated loading.

Building on the validated material models, the methodology was extended to full scale virtual prototyping of frame–sealant–glass assemblies subjected to seismic loading. A systematic design of experiments (DOE) quantifies the influence of key geometric and material parameters on critical structural responses. Together, this material to system workflow illustrates the value of advanced simulation in reducing reliance on costly, large scale physical testing, while enabling fast-to-market products.
 

 


Presenter: Hang Shu

Senior Research Specialist - Dow Chemical

Hang Shu is a Senior Research Specialist in Core R&D at THE DOW CHEMICAL COMPANY, based in Midland, Michigan. He received his PhD in Mechanical Engineering and Applied Mechanics from the University of Pennsylvania in 2024. His current role focuses on advanced material modeling and simulation of soft materials. Hang has led multiple projects supporting Dow’s Silicone and Plastics businesses, leveraging integrated experimental–computational approaches to enable predictive modeling and accelerate product development.