An octree-based adaptive mesh refinement method for three-dimensional modeling of keyhole mode laser welding. (December 2017)

• Record Type: Journal Article
• Authors: Hu, Renzhi; Pang, Shengyong; Chen, Xin; Liang, Lvjie; Shao, Xinyu
• Is Part Of: International journal of heat and mass transfer. Volume 115(2017)Part A
• Page Start: 258
• Page End: 263
• Publication Date: 2017-12
• Subjects: Octree -- Adaptive mesh refinement -- Laser welding -- High efficiency -- Spatter; Heat -- Transmission -- Periodicals; Mass transfer -- Periodicals; Chaleur -- Transmission -- Périodiques; Transfert de masse -- Périodiques; Electronic journals; 621.4022
• DOI: 10.1016/j.ijheatmasstransfer.2017.07.061
• Languages: English
• ISSNs: 0017-9310

Abstract

Highlights: A novel octree-based adaptive mesh refinement method is proposed to efficiently predict keyhole instability and weld pool dynamics of laser welding. The liquid jet can be produced when the fluid speed near the top surface of the weld pool is bigger than 1 m/s. The time for breakup of a liquid jet into spatters is around 0.065 ms. Abstract: Numerical simulations of laser welding could provide unprecedented insight into the physical details of the welding process; however, it remains a central challenge to understand laser welding with a high-efficiency and high-resolution method. This paper reports an octree-based adaptive mesh refinement method for efficient process modeling of laser welding in three dimensions. Based on multiresolution analysis algorithm, we also propose a hybrid adaptive mesh refinement strategy depending on local temperature, fluid velocity, and volume of fluid (VOF) value of weld pool of laser welding. Using the proposed method, the predicted heat transfer, and fluid flow behaviors of weld pool are consistent with experimental and previous theoretical results. Spatter formation at the beginning stage of laser welding is observed with a grid resolution in tens of microns. Under certain circumstances, the liquid jet can be produced when the fluid speed near the top surface of the weld pool is 1 m/s. The time for breakup of a liquid jet into spatters is about 0.065 ms. Furthermore, using the method, we just need 7 hours to complete a simulation of a 40-ms real laser welding process on a small workstation (2.60 GHz CPU, 16 cores), comparing with 120 hours needed using uniform grids. In a word, this method shows great potential for efficient computer simulation-based process optimization of laser material processing.