Skip to main navigation Skip to search Skip to main content

Numerical study of laser parameters on the instability of the keyhole during laser welding

  • Saeid Saediardahaei

Student thesis: Doctoral thesisDoctorate in Engineering: Engineering

Abstract

Keyhole mode laser welding is capable of achieving deep penetration in metals like aluminum by forming a vapor-induced cavity (“keyhole”) during welding. However, this process is highly susceptible to keyhole instabilities and fluctuations that can lead to defects (e.g., porosity and spatter), undermining weld quality. Direct observation and control of the keyhole are challenging due to intense optical emissions and rapid dynamics, so numerical simulation offers a valuable alternative to investigate these phenomena. This thesis aims to numerically examine how laser welding parameters affect keyhole stability in aluminum. The goal is to identify which parameters (laser power, spot size, pulse frequency/duration, waveform shape, etc.) exacerbate or mitigate keyhole instability, and to propose parameter modulation strategies (especially via power waveform shaping) to stabilize the keyhole and improve overall weld quality. A comprehensive multiphysics simulation model was developed in COMSOL Multiphysics, integrating a modified mixture theory, a level-set method for tracking the vapor/liquid and molten pool free surface, and an enthalpy-porosity approach for phase change. Using this model, both continuous wave and power-modulated pulsed laser spot welding scenarios were analyzed. Parametric studies varied laser power (2–6 kW), beam spot radius, pulse frequency, pulse duration, number of pulses, and pulse shape (rectangular, triangular, sinusoidal, cosinusoidal, etc.). Additionally, a Taguchi design of experiments combined with Grey Relational Analysis was employed to evaluate multiple weld-quality responses and determine optimal parameter combinations for stability and penetration. The simulations reveal that increasing laser intensity (higher power and longer pulses) and using a smaller beam spot greatly improve penetration (up to ~80% deeper welds) but also intensify keyhole oscillations and instability if pulses are too prolonged or frequent. In particular, higher pulse frequencies and rapid on-off cycling of the laser amplify melt pool fluctuations and increase the tendency for keyhole collapse. Force analysis showed that surface tension effects (due to keyhole curvature) and mushy-zone drag (Darcy damping) dominate the instability during pulse wave modulation, negatively affecting the balance with recoil pressure, leading to keyhole oscillations. Importantly, smoother power modulation proved effective in stabilizing the keyhole: rectangular pulses achieved the deepest penetration (exceeding that of continuous wave under equivalent conditions), whereas triangular and rampdown pulse shapes produced a more stable, cylindrical keyhole with reduced fluctuations (at the cost of slightly shallower penetration). Furthermore, systematic waveform optimization indicated that low pulse amplitude (~1000 W) and low-to-medium pulse frequencies (≈100– 400 Hz), combined with a cosinusoidal power profile, significantly improve process stability while maintaining adequate penetration. Among the parameters studied, pulse amplitude was found to have the strongest effect on weld quality (contributing over 50% of performance variation), followed by pulse frequency and waveform shape. This numerical study provides new insights into the instability of the keyhole during laser welding and demonstrates that tailored laser parameter modulation can substantially enhance welding stability. By selecting appropriate power waveforms and moderate operating conditions (avoiding extreme peak powers and overly high frequencies), it is possible to achieve deep, high-quality aluminum welds with minimal defect formation. The findings offer practical guidelines for optimizing laser welding parameters and underscore the value of multiphysics simulations in unraveling complex laser–material interaction dynamics.
Date14 Aug 2025
Original languageAmerican English
Awarding Institution
  • École de technologie supérieure
SupervisorTan Pham (Supervisor)

Cite this

'