Abstract:Computational fluid dynamics (CFD) can be a powerful tool for higher-fidelity water infrastructure planning and design. Despite decades of development and demonstration over a wide range of water systems such as clarification basins, activated sludge processes, ozone contactors, etc., CFD remains primarily used in academic research, with limited application in civil and environmental engineering practice. This limitation is contributed by its higher computational cost and demand for specialized user skills. This, however, need not be the case, if a robust and efficient surrogate model can be developed from CFD simulations and independently deployed for engineering purposes. Leveraging the emerging scientific machine learning (ML) techniques of physics-informed ML and operator learning, this study develops a composite neural network (CPNN) for learning the flow hydrodynamics and particulate matter (PM) transport and fate in clarification systems. The CPNN consists of a deep operator network (DeepONet) as an encoder and a physics-informed neural network (PINN) as a decoder. In contrast to common "black box" and lumped ML approaches, the developed CPNN directly incorporates physics principles into its architecture. Furthermore, the CPNN is designed for process-resolved and operator learning, enabling it to predict spatial hydrodynamics and PM concentration distribution (i.e., contours) across different basin geometrics and loading conditions. Compared to CFD simulation, the developed CPNN model has significantly higher computational efficiency ( ∼ milliseconds) while demonstrating robust predictive capability. For predicting basin hydrodynamics across 10,000 test cases, the trained CPNN model achieves a R2 above 0.8 for 66.4% of cases and R2 above 0.4 for 89.2% of cases. A similar performance is also demonstrated by the CPNN in predicting basin PM concentration. Further investigation reveals that basin geometrics that trigger bi-modal flow solutions can be particularly challenging for ML. Additionally, this study visualizes the dependency of basin hydrodynamics and PM concentration on basin geometrics and loading conditions, providing valuable insights for optimizing basin configuration. Lastly, the potentials and benefits of web-based applications, e.g., DeepXtorm , as a user-friendly interface for the developed CPNN model is discussed. This study represents the initial step toward achieving real-time higher-fidelity water infrastructure planning, design, optimization, and regulation.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to use Physics - Informed Neural Networks (PINNs) and operator learning techniques to develop an efficient and accurate Composite Neural Network (CPNN) to predict the hydrodynamic and particulate matter transport processes in urban water clarification systems. Specifically, the objectives of the paper are as follows: 1. **Improve computational efficiency**: Compared with traditional Computational Fluid Dynamics (CFD) simulations, the CPNN model can significantly reduce computational costs (reaching millisecond - level computation time) while maintaining high prediction accuracy. This is of great significance for large - scale engineering applications, such as the design and optimization of water treatment facilities. 2. **Enhance the generalization ability of the model**: By combining the Deep Operator Network (DeepONet) as an encoder and the Physics - Informed Neural Network (PINN) as a decoder, the CPNN can directly incorporate physical principles into its architecture, thereby better handling water clarification systems under different geometries and loading conditions. This enables the model to not only predict results under specific conditions but also provide reliable predictions under a wide range of conditions. 3. **Achieve real - time design and optimization**: The ultimate goal is to use the CPNN model to achieve real - time high - fidelity operations in water infrastructure planning, design, optimization, and supervision. This will help improve the efficiency and cost - effectiveness of water treatment systems while reducing the dependence on high - performance computing resources. ### Specific research contents of the paper - **Methodology**: - **Definition of operator learning tasks**: The prediction problems of hydrodynamic and particulate matter transport in water clarification systems are formalized as an operator learning task, where the input parameters include basin geometry and loading conditions, and the output is the flow field and particulate matter concentration distribution. - **High - fidelity CFD simulation**: Use a validated CFD model (basinFoam) to generate a large amount of training data under different geometries and loading conditions. - **Construction of Composite Neural Network (CPNN)**: Combine the advantages of DeepONet and PINN to construct a composite neural network that can efficiently learn and predict complex hydrodynamic and particulate matter transport processes. - **Experimental results**: - **Prediction performance**: Among 10,000 test cases, when the trained CPNN model predicts the basin hydrodynamic, the R² value is higher than 0.8 in 66.4% of the cases and higher than 0.4 in 89.2% of the cases. Similar good performance is also shown in the prediction of particulate matter concentration. - **Model robustness**: Further research shows that some geometries (such as basins that cause bimodal flow) are more challenging for machine learning models. - **Visual analysis**: Through visual analysis, the influence of basin geometry and loading conditions on hydrodynamic and particulate matter concentration is shown, providing valuable insights for optimizing basin configuration. ### Conclusion By combining Physics - Informed Neural Networks and operator learning techniques, the CPNN model performs excellently in predicting the hydrodynamic and particulate matter transport in urban water clarification systems, with high computational efficiency and good prediction ability. This provides new possibilities for achieving real - time high - fidelity design and optimization of water infrastructure.

Operator learning for urban water clarification hydrodynamics and particulate matter transport with physics-informed neural networks

Unit Operation and Process Modeling with Physics-Informed Machine Learning

Implementing Machine Learning to Optimize the Cost-Benefit of Urban Water Clarifier Geometrics.

Physics-enhanced Neural Operator for Simulating Turbulent Transport

Modelling clogging dynamics in groundwater systems using multiscale homogenized physics informed neural network (MHPINN)

Surrogate Modeling for Unsaturated Infiltration Via the Physics and Equality-Constrained Artificial Neural Networks

Learning Generic Solutions for Multiphase Transport in Porous Media via the Flux Functions Operator

Investigating Steady Unconfined Groundwater Flow using Physics Informed Neural Networks

Enhanced physics‐informed neural networks for efficient modelling of hydrodynamics in river networks

Physics-Informed Neural Networks for Modeling Water Flows in a River Channel

Bridging Hydrological Ensemble Simulation and Learning Using Deep Neural Operators

A physics-informed convolutional neural network for the simulation and prediction of two-phase Darcy flows in heterogeneous porous media

Modeling Transient Mixed Flows in Sewer Systems with Data Fusion Via Physics-Informed Machine Learning

A CFD-ML augmented alternative to residence time for clarification basin scaling and design

Porous-DeepONet: Learning the Solution Operators of Parametric Reactive Transport Equations in Porous Media

Physics-informed neural networks of the Saint-Venant equations for downscaling a large-scale river model

Learning Groundwater Contaminant Diffusion‐Sorption Processes with a Finite Volume Neural Network

Machine Learning Parallel System for Integrated Process-Model Calibration and Accuracy Enhancement in Sewer-River System

Application of Deep Neural Network Technology for Multi‐scale CFD Modeling in Porous Media

Interfacing finite elements with deep neural operators for fast multiscale modeling of mechanics problems

Physics-informed neural networks for modeling two-phase steady state flow with capillary heterogeneity at varying flow conditions