skip to main content

Mechanical and Civil Engineering Seminar

Thursday, October 10, 2024
11:00am to 12:00pm
Add to Cal
Gates-Thomas 135
An Interleaved Physics-Based Deep-Learning Framework as a New Cycle Jumping Approach for Microstructurally Small Fatigue Crack Growth Simulations
Ashley Spear, Associate Professor, Department of Mechanical Engineering, University of Utah,

Mechanical and Civil Engineering Seminar Series

Title: An Interleaved Physics-Based Deep-Learning Framework as a New Cycle Jumping Approach for Microstructurally Small Fatigue Crack Growth Simulations

Abstract: The early stages of fatigue-crack evolution can consume the majority of a structure's life in high-cycle fatigue applications. Thus, predicting accurately the behavior of microstructurally small fatigue cracks (MSCs) is essential for developing next-generation fatigue-resistant materials and for realizing concepts like the airframe digital twin. While current simulation frameworks using crystal plasticity constitutive models can resolve material deformation and micromechanical fields at the MSC scale, running such simulations over realistic cycle counts remains computationally intractable. This work introduces a novel cycle-jumping strategy that leverages deep learning and uncertainty quantification (UQ) to accelerate 3D MSC propagation simulations. Bidirectional Long Short-Term Memory (BiLSTM) networks are trained to predict local crack deflection and growth rate using 18,000 data sequences extracted from 40 high-fidelity, physics-based simulations of MSC propagation. Recognizing that making many successive forward predictions using the BiLSTM deep-learning framework can lead to unacceptable uncertainty propagation, we propose an interleaved physics-based deep-learning (PBDL) framework that combines the rapid predictive capabilities of deep learning with the accuracy of physics-based models. In the proposed framework, UQ plays a key role in determining when to update the explicit crack surface in the physics-based model with the deep-learning evolved crack surface, prior to resuming deep-learning predictions using the updated physics-based model for input. The UQ-informed PBDL framework enables the simulation of MSC growth over a realistic number of cycle counts while reining in model error and uncertainty propagation. The work represents a significant advancement in fatigue modeling and offers a template for other applications.

Bio: Dr. Ashley Spear is a Presidential Scholar at the University of Utah and an Associate Professor in Mechanical Engineering. She directs the Multiscale Mechanics & Materials Laboratory, which specializes in integrating physics-based modeling, data science, and experiments to examine deformation, fatigue, and fracture in a wide range of materials. Spear received her B.S. in Architectural Engineering from the University of Wyoming and Ph.D. in Civil Engineering from Cornell University. She is the recipient of the Constance Tipper Medal from the International Congress on Fracture, the Young Investigator Award from the Air Force Office of Scientific Research, the TMS Early Career Faculty Fellow Award, and the National Science Foundation CAREER award.

NOTE: At this time, in-person Mechanical and Civil Engineering Lectures are open to all Caltech students/staff/faculty/visitors.

For more information, please contact Kristen Bazua by phone at (626) 395-3385 or by email at [email protected] or visit https://www.mce.caltech.edu/seminars.