Digital Twin of an Argon-Hydrogen Plasma Reactor

Abstract

The principal proof of concept revolves around an argon-hydrogen plasma reactor that melts, reduces, atomizes and quenches critical raw material in one step, with premium spherical powder as the deliverable output and control of the composition chemistry. Each usage of the reactor is monitored through thermocouples and pressure sensors, which provide a daily data source of the real-world experiments. The reactor is modeled through COMSOL Multiphysics, which represents the core solver used to provide multiphysics simulations. The usage of COMSOL is complemented with Artificial Intelligence (AI) models, to enable seamless data assimilation and optimization. This paper presents the COMSOL twin of the reaction chamber and converging-diverging nozzle, together with a custom phase-change particle-tracing layer validated on Ti-6Al-4V (Ti64). Moreover, we highlight how the synergy between COMSOL simulations and AI-based digital surrogates can be leveraged to build self-consistent optimization loops geared toward (i) fully autonomous live control of the reactor and (ii) optimization of the process.