Dynamic beam shaping enables diverse possibilities to influence the performance of laser-based manufacturing processes, for example by tailoring the spatial intensity distribution of light in real time. One such possibility is shape sequencing, which involves switching between two or more beam shapes during laser-matter interaction, with each shape being active for a specified time frame (shape duration) within the sequence. When properly understood and utilized, shape sequencing could serve as an effective optimization approach in high-power laser materials processing applications such as Laser Beam Welding (LBW). In this work, a multi-physical simulation model is used as a comprehensive framework to evaluate the impact of shape duration on the dynamics of the LBW process, with the aim of mitigating instabilities that can lead to defects such as pores, spatter, lack of penetration, cracks, etc. The long-term goal is not only to identify optimal welding parameters, but also to understand why certain shaping strategies lead to better or worse process outcomes. As a demonstration, an aluminum-magnesium alloy is chosen as the workpiece material for a deep penetration welding (DPW) scenario. The numerical model is initially validated by comparing simulation results with macrographs of experimentally produced welds and is subsequently used to analyze a welding setup incorporating dynamic beam shape sequencing. The study reveals that shape sequencing can effectively induce vibrations at predefined frequencies on the keyhole surface, manifesting as periodic pulses whose width is determined by the shape durations. This pulsed excitation is subsequently transmitted to the melt pool, influencing its behavior. Certain shape durations favor a more or less effective excitation of the melt pool.