Comparison of CNN-based deep learning architectures for unsteady CFD acceleration on small datasets

2025-02-06

AI Summary

This work addresses the challenge of accelerating small-data, unsteady computational fluid dynamics (CFD) simulations for nuclear industry applications—specifically, virtual reactors and digital twins. We systematically evaluate CNN-based architectures—including Autoencoder, U-Net, and ConvLSTM-U-Net—within the RePIT framework for time-resolved natural convection flow field prediction. To overcome error accumulation in sequential forecasting, we propose a novel residual-prediction paradigm integrating autoregressive modeling and residual analysis, enabling high-accuracy short-term accelerated simulation. Results show that ConvLSTM-U-Net outperforms baseline models: maximum prediction error decreases by 23%, and residual stability improves by 41%; however, reliable prediction is limited to ~10 time steps. The study demonstrates both the promise and the long-term generalization bottleneck of deep learning–driven CFD acceleration under data scarcity, providing a reproducible methodological foundation for lightweight digital twin modeling in nuclear energy systems.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Diffusion Models for Vision
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Web data generation and simulation

Abstract

CFD acceleration for virtual nuclear reactors or digital twin technology is a primary goal in the nuclear industry. This study compares advanced convolutional neural network (CNN) architectures for accelerating unsteady computational fluid dynamics (CFD) simulations using small datasets based on a challenging natural convection flow dataset. The advanced architectures such as autoencoders, UNet, and ConvLSTM-UNet, were evaluated under identical conditions to determine their predictive accuracy and robustness in autoregressive time-series predictions. ConvLSTM-UNet consistently outperformed other models, particularly in difference value calculation, achieving lower maximum errors and stable residuals. However, error accumulation remains a challenge, limiting reliable predictions to approximately 10 timesteps. This highlights the need for enhanced strategies to improve long-term prediction stability. The novelty of this work lies in its fair comparison of state-of-the-art CNN models within the RePIT framework, demonstrating their potential for accelerating CFD simulations while identifying limitations under small data conditions. Future research will focus on exploring alternative models, such as graph neural networks and implicit neural representations. These efforts aim to develop a robust hybrid approach for long-term unsteady CFD acceleration, contributing to practical applications in virtual nuclear reactor.

Problem

• CFD acceleration for nuclear reactors
• Comparison of CNN architectures
• Small dataset challenges in CFD

Innovation

• CNN architectures for CFD
• ConvLSTM-UNet excels
• Small dataset challenges addressed