Data driven model for grain structure evolution of the powder bed fusion on austenitic stainless steel using machine learning

The powder bed fusion (PBF) process is a type of additive manufacturing that involves the use of laser or electron beam to selectively melt and fuse layers of metallic powder, building up a three-dimensional object. The microstructural properties of the final product are critical to its perfor...

Full description

Saved in:
Bibliographic Details
Main Author: Feng, Cheng
Other Authors: Yeong Wai Yee
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2023
Subjects:
Online Access:https://hdl.handle.net/10356/167207
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:The powder bed fusion (PBF) process is a type of additive manufacturing that involves the use of laser or electron beam to selectively melt and fuse layers of metallic powder, building up a three-dimensional object. The microstructural properties of the final product are critical to its performance, and are influenced by a variety of process parameters, such as laser power, scan speed, and powder bed temperature. However, simulating the microstructure of PBF parts is computationally intensive and requires significant time and resources, which can be a challenging for designers and engineers who need to quickly optimize the design and production process. A proposed solution is to produce a machine learning model that can predict the evolution of the microstructure, which can save material and costs. For this project, 3D-CNN LSTM model was used for the machine learning, by using the inputs of OPENFOM simulation data and generated GIF files from PARAVIEW as the dataset. Although the model has shown high accuracy in 2-dimensional prediction, it has limitations in capturing the cross-sectional view of the melt pool in the microstructure, hindering a comprehensive investigation, Structural Similarity Index and pixel intensity method were used for the data analysis part and the outcome is stated below. Therefore, further research can focus on developing a machine learning model to predict the microstructure evolution in 3-dimensional. By using ML to predict the microstructure of parts based on their process parameters, designers and engineers can quickly explore a wide range of design options and process parameters, without the need for costly and time-consuming. This can lead to significant reductions in the time and cost required to develop new PBF parts, while also improving their performance and quality.