Prediction of Temperature on the Substrate During Wire Arc DED Through Physics Informed Neural Network

DOI

10.2991/978-94-6239-813-9_7How to use a DOI?

Keywords

Additive manufacturing; PINN; Wire arc DED; Substrate temperature

Abstract

Accurate prediction of the substrate temperature during Wire Arc Directed Energy Deposition (Wire Arc DED) is critical for managing thermal gradients, avoiding residual deformation, and maintaining the dimensional correctness and mechanical performance of manufactured metallic components. The research introduces a new physics-informed neural network (PINN) framework designed for predicting a time-dependent temperature distribution in the substrate using only the data obtained by numerical simulations. To start with, a three-dimensional heat model in transient mode was created using COMSOL Multiphysics to study heat transfer processes throughout deposition. The temperature data obtained from four selected areas on the substrate were used to train the PINN system based on the theory of heat transfer used in WA-DED. It was proven that the proposed model was able to produce accurate predictions when tested against finite element modelling data. additionally, the effect of the main process factors on the temperature in the substrate was analysed. In addition, comparative analysis revealed that the PINN model consistently outperformed a conventional data-driven Deep Neural Network (DNN), achieving lower values of Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and MAPE, thereby confirming its superior accuracy, robustness, and generalization capability. The proposed PINN framework offers a computationally efficient and physics-consistent approach for thermal field prediction, providing a promising tool for real-time thermal monitoring, process optimization, and intelligent control of WA-DED manufacturing.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Rupendra Singh Tanwar
AU  - Suyog Jhavar
AU  - Sheetal Kumar Dewangan
PY  - 2026
DA  - 2026/10/29
TI  - Prediction of Temperature on the Substrate During Wire Arc DED Through Physics Informed Neural Network
BT  - Proceedings of the  Global Symposium on Adaptive Manufacturing (GSAM 2026)
PB  - Atlantis Press
SP  - 19
EP  - 21
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6239-813-9_7
DO  - 10.2991/978-94-6239-813-9_7
ID  - Tanwar2026
ER  -