Physics-Informed and Operator Learning Models for Cavitation in Ultrasonic Cleaning Systems

• Applicant: Dr.-Ing. Yu Jiao
• Subject Area: Fluid Mechanics
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 576724787

The central objective of this project is to develop surrogate models that enable fast, reliable, and physically consistent predictions of key performance indicators of ultrasonic cleaning, particularly acoustic field distributions and surface pressure loadings near the components to be cleaned. For example, by specifying input parameters such as acoustic intensity, excitation frequency, transducer configuration, and component placement, the surrogate models will provide rapid predictions of the resulting acoustic field and surface pressure loadings. These outputs serve as critical metrics for evaluating cleaning effectiveness while reducing computational costs by orders in magnitude compared to fully resolved simulations. Ultimately, the surrogate model will enable reverse design of transducer configurations and key operating parameters, such as excitation frequency and placement, contributing to improved cleaning performance and broader industrial applicability. Specifically, this project is structured around three core parts: high-fidelity simulation of acoustic and bubble dynamics within a reduced parameter space (WP1); surrogate modeling for real-time prediction of acoustic fields and bubble dynamics (WP2&3); and process optimization and practical recommendations (WP4). WP stands for 'work package'.

• DFG Programme: WBP Position