![[2602.22188] Surrogate models for Rock-Fluid Interaction: A Grid-Size-Invariant Approach](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22188] Surrogate models for Rock-Fluid Interaction: A Grid-Size-Invariant Approach
Summary
This article presents a novel approach to modeling rock-fluid interactions using surrogate models that are grid-size invariant, enhancing computational efficiency and predictive accuracy.
Why It Matters
The development of surrogate models for rock-fluid interactions is crucial for applications requiring rapid simulations, such as uncertainty quantification and optimization. This research addresses the limitations of traditional high-fidelity models, making advanced modeling techniques more accessible and efficient for various scientific and engineering applications.
Key Takeaways
- Introduces eight surrogate models for predicting fluid flow in porous media.
- Develops grid-size-invariant models that reduce memory consumption during training.
- Demonstrates that UNet++ architecture outperforms UNet in predictive performance.
- Addresses challenges posed by fluid-induced rock dissolution in modeling.
- Enhances the applicability of models for multi-query problems.
Computer Science > Machine Learning arXiv:2602.22188 (cs) [Submitted on 25 Feb 2026] Title:Surrogate models for Rock-Fluid Interaction: A Grid-Size-Invariant Approach Authors:Nathalie C. Pinheiro, Donghu Guo, Hannah P. Menke, Aniket C. Joshi, Claire E. Heaney, Ahmed H. ElSheikh, Christopher C. Pain View a PDF of the paper titled Surrogate models for Rock-Fluid Interaction: A Grid-Size-Invariant Approach, by Nathalie C. Pinheiro and 6 other authors View PDF HTML (experimental) Abstract:Modelling rock-fluid interaction requires solving a set of partial differential equations (PDEs) to predict the flow behaviour and the reactions of the fluid with the rock on the interfaces. Conventional high-fidelity numerical models require a high resolution to obtain reliable results, resulting in huge computational expense. This restricts the applicability of these models for multi-query problems, such as uncertainty quantification and optimisation, which require running numerous scenarios. As a cheaper alternative to high-fidelity models, this work develops eight surrogate models for predicting the fluid flow in porous media. Four of these are reduced-order models (ROM) based on one neural network for compression and another for prediction. The other four are single neural networks with the property of grid-size invariance; a term which we use to refer to image-to-image models that are capable of inferring on computational domains that are larger than those used during training. In addit...