Physical Understanding of Active Control of Beam Scanning in Preventing Top Concavity in Electron Beam Welding

Abstract

Low-frequency scanning electron beam welding (SEBW) is recently recognized as a possible control method for preventing the top concavity behavior in plate full penetration welding. However, due to a lack of understanding of the control of beam scanning in preventing concavity, the welding quality is difficult to control. In this study, the basic physics under different scanning frequencies was investigated. The results indicate the fundamental reason of improving the weld morphology is that the scanning can make the energy more dispersed. Specifically, due to the impact of recoil pressure and the nature of the weld pool, longitudinal scanning can promote the upstream flow of liquid metal and prevent the top concavity. However, when the scanning frequency reaches a certain threshold value, the effect of promoting the upwelling becomes weak. The threshold of the scanning frequency is 42.00 Hz for welding the 22-mm-thick 30HGSA steel in this study. When the scanning frequency is between 24.00 and 42.00 Hz, the process is in the Improvement Stage. Therefore, longitudinal scanning can effectively control the flow in the weld pool, and an appropriate scanning frequency is necessary to prevent the top concavity. These investigations can provide a better understanding of the effect of scanning on the top morphology.

Graphical abstract

Appendices

Appendix A

A.1 Simulation Methods

To simplify the energy interaction process between the high energy beam and the transient change keyhole surface, the Dirac delta function was adopted,^[23] which is designed to be convenient for Level Set (LS) method.^[24] In addition, a ray casting method^[25] was adopted to consider the situation that some parts of the keyhole surface were blocked by the protrusions on the upper of them.

To create the physical process of phase transition between the solid and the liquid, and the forces like interphase force, interphase force, and so on during EBW, the mixture model presented by Eggers was adopted,^[26] assuming the fluid is incompressible

As for the motion description of the keyhole free surface, LS method^[24] was adopted to track the evolution process.

Discontinuous boundary model proposed by Pang et al. was adopted to treat the boundary conditions in SEBW, considering the discontinuity caused by Marangoni shear stress, surface tension, and recoil pressure.^[20] The model was reviewed by Tang et al.,^[27] and a specific description of the discretization process can also be found in Pang et al. by adopting the high-order finite difference method.^[28] Most of the physical parameters adopted in the process were determined by using the thermal parameter calculation software JMatPro. Figure

Fig. A.1

Simulation flowchart of SEBW

A.1 presents the calculating flowchart.^[29]

A.2 Modeling Verification

To prove the validity of the visualization results, the depth and average width of the Sample No. 1 are listed in Table

Table A.1 Weld Section Dimension Comparison of the Simulated Result with the Experimental Result About Sample No. 1.

A.1. Furthermore, the weld section profile calculated was compared against the cross section observed in Figure

Fig. A.2

The comparison of the weld section simulated against the cross section obtained in experiment about Sample No. 1. (a) morphology of cross-sectional fusion zone; (b) profile of cross-sectional fusion zone

A.2. The upper part of Figure A.2(a) is the simulation result, and the lower part is the experimental result. To clearly exhibiting the difference, a pair of comparable contour profiles are extracted from Figure A.2(a), which represents the fusion zone of the cross section obtained in two different ways, as is shown in Figure A.2(b). According to Figure A.2, a similar result is presented that the profiles of the cross section between the experimental result and the simulated result are within the margins of error. Still, a visible difference appears at the opening of the keyhole leading to air, indicating the limitation of the model that it sacrifices the accurate estimate of vapor plume friction for shortening the computing time.