Advancing Digital Twins of Wire Arc-DED Through Process Control with a Multi-Modal Sensor Array

Betts, J. L. (2024). Advancing Digital Twins of Wire Arc-DED Through Process Control with a Multi-Modal Sensor Array. Mississippi State University: Proquest. 259.

Abstract

The goal of this work was to advance digital twins of wire-arc directed energy deposition (arc-DED) through process control with a multi-modal sensor array. Digital twins allow for synchronization, context, and visualization of in situ data. Additionally digital twins allow for bi-directional communication between the virtual and physical system, allowing for process control.This work implemented feedback control of the contact tip to work piece distance (CTWD), established a standard method to create modular unified robot description format (URDF) of the arc-DED system, and produced a complex component while collecting multi-modal data within the digital twin for robotic additive welding (DRAW) powered by Robot Operating System 2 (ROS2). This work also developed a robust pre- and post-processing framework in Python for three-dimensional FE thermal models of arc-DED. The framework looked at the effects of processing parameters and model convergence, through a full factorial design of experiments. Controlling CTWD is critical for wire-arc DED, and this work implemented a novel method of feedback control that can measure the weld bead height in situ. It is imperative for digital twins to have a comprehensive and accurate virtual representation. The standard method to create a URDF for arc-DED outlined in this work includes the entire robotic welding cell, all robotic axes, and sensors in an accurate, and modular framework. Following this development a large complex geometry (42.78 lb) using arc-DED, and DRAW reliably captured approximately 200 Gb of multi-modal data over the twenty-two hours of manufacturing time. This demonstrated the ability to produce a forty-two-pound complex component, four times larger than the next largest component produced on this system prior. Finally, this demonstrated the reliability of utilizing ROS 2 for multi-modal data capture and process control.