Feature-Modulated UFNO for Improved Prediction of Multiphase Flow in Porous Media

2025-11-25

AI Summary

To address the spectral redundancy arising from forcing scalar inputs (e.g., temperature, injection rate) into spatial fields in the Universal Fourier Neural Operator (UFNO) for multiphase flow prediction in porous media—and the insensitivity of standard loss functions to spatially varying error significance—this paper proposes UFNO-FiLM. Our method introduces Feature-wise Linear Modulation (FiLM) layers to decouple and conditionally modulate spatial features with scalar inputs, and designs a physics-informed spatially weighted loss function that emphasizes high-importance regions. Integrating Fourier neural operators with a U-Net architecture, UFNO-FiLM performs conditional modeling at the feature level. Experiments on subsurface multiphase flow prediction demonstrate that UFNO-FiLM reduces gas saturation MAE by 21% over UFNO, significantly improving predictive accuracy and generalization—particularly in physically critical regions such as fluid fronts and wellbore vicinities.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Search and Optimization: Learning to Search
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

The UNet-enhanced Fourier Neural Operator (UFNO) extends the Fourier Neural Operator (FNO) by incorporating a parallel UNet pathway, enabling the retention of both high- and low-frequency components. While UFNO improves predictive accuracy over FNO, it inefficiently treats scalar inputs (e.g., temperature, injection rate) as spatially distributed fields by duplicating their values across the domain. This forces the model to process redundant constant signals within the frequency domain. Additionally, its standard loss function does not account for spatial variations in error sensitivity, limiting performance in regions of high physical importance. We introduce UFNO-FiLM, an enhanced architecture that incorporates two key innovations. First, we decouple scalar inputs from spatial features using a Feature-wise Linear Modulation (FiLM) layer, allowing the model to modulate spatial feature maps without introducing constant signals into the Fourier transform. Second, we employ a spatially weighted loss function that prioritizes learning in critical regions. Our experiments on subsurface multiphase flow demonstrate a 21% reduction in gas saturation Mean Absolute Error (MAE) compared to UFNO, highlighting the effectiveness of our approach in improving predictive accuracy.

Problem

• Inefficient handling of scalar inputs as constant spatial fields
• Standard loss function ignores spatial error sensitivity variations
• Limited performance in critical regions of multiphase flow

Innovation

• FiLM layer decouples scalar inputs from spatial features
• Spatially weighted loss prioritizes critical region learning
• Modulates spatial maps without constant Fourier signals