Physics-based multiphysics modeling for process tailoring in ultrasonic welding of thermoplastic composites

This study presents a computationally efficient sequentially coupled thermo-mechanical modeling framework for predicting transient temperature evolution and enabling physics-based process tailoring in ultrasonic welding of thermoplastic composites. Unlike conventional one-way sequential or fully coupled approaches, the developed modeling strategy iteratively exchanges information between dynamic mechanical and transient heat-transfer analyses throughout the welding process, allowing the evolving temperature field to continuously update the mechanical response and associated heat generation. Viscoelastic and frictional heating are computed from the dynamic response of the composite adherends and energy director under ultrasonic excitation, using temperature-dependent thermal and mechanical properties. Temperature-dependent material properties were obtained from the literature, and a single hammering factor was calibrated using welding experiments at a vibration amplitude of 70 μm amplitude. Model predictions were validated against infrared thermography from experiments conducted at 80 μm and 90 μm amplitudes, showing close agreement in both magnitude and spatial/temporal trends of the temperature field, with prediction errors generally falling within the standard deviation of the experimental measurements. In addition, the simulated welded area closely matched the experimentally measured post-weld area, with molten interface fractions of 23% versus 29%, 87% versus 100%, and 94% versus 100% for the 70, 80, and 90 μm cases, respectively, demonstrating the model’s capability to establish quantitative relationships among process conditions, material behavior, temperature evolution, and weld quality. Finally, the validated model was used to determine an optimal welding time that improves weld quality and process efficiency. Implementing this predicted optimum in subsequent experiments resulted in reduced fiber distortion, a high welded fraction of 91.7%, compared to 92.3% predicted by simulation, and a 20% reduction in processing time, confirming the value of the proposed framework for efficient, physics-based digital process-tailoring of ultrasonic welding.