Cross-Field Interface-Aware Neural Operators for Multiphase Flow Simulation

TL;DR

The paper tackles the computational and data-efficiency challenges of simulating multiphase flows by introducing the Interface Information Aware Neural Operator (IANO). It combines an interface-aware function encoding with a geometry-aware positional encoding to capture cross-field interactions and spatial correlations at phase interfaces, addressing spectral bias and data scarcity. Empirical results show up to ~10% accuracy gains over baselines, plus robust performance in low-data and noisy settings and strong super-resolution capabilities. The approach offers a practical path toward data-efficient, AI-based multiphase flow simulations that leverage readily available interface data.

Abstract

Multiphase flow simulation is critical in science and engineering but incurs high computational costs due to complex field discontinuities and the need for high-resolution numerical meshes. While Neural Operators (NOs) offer an efficient alternative for solving Partial Differential Equations (PDEs), they struggle with two core challenges unique to multiphase systems: spectral bias caused by spatial heterogeneity at phase interfaces, and the persistent scarcity of expensive, high-resolution field data. This work introduces the Interface Information Aware Neural Operator (IANO), a novel architecture that mitigates these issues by leveraging readily obtainable interface data (e.g., topology and position). Interface data inherently contains the high-frequency features not only necessary to complement the physical field data, but also help with spectral bias. IANO incorporates an interface-aware function encoding mechanism to capture dynamic coupling, and a geometry-aware positional encoding method to enhance spatial fidelity for pointwise super-resolution. Empirical results across multiple multiphase flow cases demonstrate that IANO achieves significant accuracy improvements (up to $\sim$10\%) over existing NO baselines. Furthermore, IANO exhibits superior generalization capabilities in low-data and noisy settings, confirming its utility for practical, data-efficient $\text{AI}$-based multiphase flow simulations.

Figures (11)

• Figure 1: IANO combines interface data as auxiliary variables with the multiphysics field data for multiphase flow simulation.
• Figure 2: IANO's architecture: 1) The interface-aware multiple functions encoding mechanism jointly encodes both multiphysics fields and interface information to generate an interface-aware function embedding for each field. 2) The geometry-aware positional encoding mechanism produces positional embedding by explicitly linking multiphysics fields and interfaces to their positions. IANO integrates the function embeddings with the positional embeddings for multiphase flow prediction.
• Figure 3: The RMSE error plot for Subcooled Pool Boiling at 2× resolution. IANO consistently achieves the lowest RMSE.
• Figure : (a)
• Figure : (a)