Construction of digital twin of the space furnace based on thermal simulation and data-driven thermophysical parameter identification technology

Abstract

This study aims to demonstrate the construction of a simulation-based digital twin of a space furnace. During simulation processes, uncertainties in the results can arise from variations in parameter inputs and the utilization of different simulation models. Moreover, simulating complex models often demands substantial computational resources. To effectively achieve the desired digital twin effect, it is crucial to employ specific measures that address these challenges. In this paper, we propose a layered meshing strategy that aims to strike a balance between simulation accuracy and computational costs. By implementing this strategy, we can optimize the mesh design and achieve accurate results while efficiently managing computational resources. The performance of the meshing strategy is evaluated based on three indices: orthogonal quality, skewness, and aspect ratio. To enhance the repeatability of simulation results, we employ two key methods: model selection and thermophysical parameter identification. Laminar and discrete ordinate (DO) models are selected by comparing simulation results from different model combinations. A data-driven method is proposed to identify the thermophysical parameters when prior knowledge is lacking. To assess the validity of the proposed method, simulation results are compared with state-of-art methods. Results show satisfactory agreement between measured temperatures and simulated temperatures, with the relative error being approximately 1% at the temperature control thermocouples. The proposed thermophysical identification method achieves equivalent or better error performance compared to cumbersome manual adjustment and instrument measurement methods, which reduces experimental risks and costs.