Author ORCID Identifier
Semester
Fall
Date of Graduation
2024
Document Type
Dissertation
Degree Type
PhD
College
Statler College of Engineering and Mineral Resources
Department
Mechanical and Aerospace Engineering
Committee Chair
Eduardo M. Sosa
Committee Member
Ever J. Barbero
Committee Member
Victor H. Mucino
Committee Member
Gregory J. Thompson
Committee Member
Chao Zhang
Abstract
This research presents the development of a data-driven machine learning-assisted approach to simulate and optimize the design of composite materials subjected to low-velocity impacts. It focuses on a specific type of hybrid composite comprised of fiberglass and Kevlar fabrics stitched with Kevlar threads. The study begins by creating a multiscale finite element simulation for the composite under a low-velocity impact, which operates across three scales: microscale, mesoscale, and macroscale. This simulation accepts various inputs, such as different layer configurations and orientations, and provides impact outputs including maximum load and energy absorption capacity and displacement at failure.
Python and MATLAB scripts manage the simulation, automating data generation to create a large dataset for analyzing the composite under low-velocity impacts at both the mesoscale and macroscale levels. This data then trains artificial neural networks (ANNs) at each scale, which are coupled to form a multiscale machine learning-assisted simulation. The mesoscale ANN uses homogenized mechanical properties (HMPs) from the microscale model of the composite layers to predict mesoscale HMPs where a combination of layers and stitching threads are considered in a mesoscale representative volume element (RVE). The macroscale ANN then uses the mesoscale HMPs to produce the composite's impact response, including maximum force and energy absorption capacity and displacement at failure.
Finally, the multiscale ANN system is integrated with a particle swarm optimization (PSO) algorithm to explore the design space and find the optimal composite configuration for specific impact conditions and constraints. All simulation results and predictions from the multiscale ANN are validated against experimental data on composites subjected to low-velocity impacts. The multiscale ANN approach introduced in this study is flexible and can be applied to other complex materials and structures where several variables across different scales influence the results. This study shows the efficiency of the introduced approach specifically for the optimization process where several possible combinations of the design space need to be evaluated to find the best combination for a given set of constraints.
Recommended Citation
Mohammadi, Masoud, "Machine learning-assisted multiscale simulation and design optimization of composite jackets under low-velocity impacts" (2024). Graduate Theses, Dissertations, and Problem Reports. 12661.
https://researchrepository.wvu.edu/etd/12661
Included in
Applied Mechanics Commons, Computer-Aided Engineering and Design Commons, Structural Materials Commons, Structures and Materials Commons
