The calculation of the curvature in Volume of Fluid (VOF) methods is still a challenge, and common approaches involve curve or surface fitting based on volume fractions. In this work, we explore an alternative approach for curvature computation in VOF simulations employing machine learning. The neural network establishes a correlation between curvature and height function values so that the local interface curvature can be efficiently predicted. We compare the trained neural network to the standard Height Function method to assess its performance and robustness.
Cervone, A., Manservisi, S., Scardovelli, R., Sirotti, L. (2025). A Machine Learning Approach for Curvature Computation Leveraging Volume of Fluid (VOF) Height Function. American Institute of Physics [10.1063/5.0286349].
A Machine Learning Approach for Curvature Computation Leveraging Volume of Fluid (VOF) Height Function
Cervone A.;Manservisi S.;Scardovelli R.;Sirotti L.
2025
Abstract
The calculation of the curvature in Volume of Fluid (VOF) methods is still a challenge, and common approaches involve curve or surface fitting based on volume fractions. In this work, we explore an alternative approach for curvature computation in VOF simulations employing machine learning. The neural network establishes a correlation between curvature and height function values so that the local interface curvature can be efficiently predicted. We compare the trained neural network to the standard Height Function method to assess its performance and robustness.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
