Abstract:The high-pressure transportation process of pipeline necessitates an accurate hydraulic transient simulation tool to prevent slack line flow and over-pressure, which can endanger pipeline operations. However, current numerical solution methods often face difficulties in balancing computational efficiency and accuracy. Additionally, few studies attempt to reform physics-informed learning architecture for pipeline transient simulation with magnitude different in outputs and imbalanced gradient in loss function. To address these challenges, a Knowledge-Inspired Hierarchical Physics-Informed Neural Network is proposed for hydraulic transient simulation of multi-product pipelines. The proposed model integrates governing equations, boundary conditions, and initial conditions into the training process to ensure consistency with physical laws. Furthermore, magnitude conversion of outputs and equivalent conversion of governing equations are implemented to enhance the training performance of the neural network. To further address the imbalanced gradient of multiple loss terms with fixed weights, a hierarchical training strategy is designed. Numerical simulations demonstrate that the proposed model outperforms state-of-the-art models and can still produce accurate simulation results under complex hydraulic transient conditions, with mean absolute percentage errors reduced by 87.8\% and 92.7 \% in pressure prediction. Thus, the proposed model can conduct accurate and effective hydraulic transient analysis, ensuring the safe operation of pipelines.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to balance computational efficiency and accuracy in hydraulic transient simulations in multi - product pipelines, and to address the issues of output magnitude differences and unbalanced loss function gradients. Specifically: 1. **Improve computational efficiency and accuracy**: Current numerical solutions often struggle to simultaneously ensure computational efficiency and the accuracy of results when dealing with pipeline hydraulic transient simulations. This limits their effectiveness in practical applications. 2. **Solve output magnitude differences**: In hydraulic transient simulations of multi - product pipelines, there are significant magnitude differences between different output variables (such as pressure and flow velocity), which can lead to poor gradient - descent performance during neural network training and affect the prediction accuracy of the model. 3. **Handle unbalanced loss function gradients**: In the training process of traditional physics - informed neural networks (PINNs), the weights of multiple loss terms are fixed, which easily leads to unbalanced gradients, thereby affecting the convergence effect of the model and the final prediction performance. To solve the above problems, the paper proposes a knowledge - inspired hierarchical physics - informed neural network (KIH - PINN), which improves the performance of the model through the following methods: - **Integrate physical laws**: Incorporate the governing equations, boundary conditions, and initial conditions into the training process to ensure that the results generated by the model conform to physical laws. - **Output magnitude conversion**: Perform magnitude conversion on output variables to make the magnitudes of different output variables tend to be consistent, thereby improving the gradient - descent performance. - **Hierarchical training strategy**: Design a hierarchical training strategy that decomposes the training process into multiple sub - processes and gradually optimizes the model parameters to overcome the gradient imbalance problem caused by the fixed - weight loss function. Through these innovations, the KIH - PINN model exhibits higher prediction accuracy and computational efficiency under complex hydraulic transient conditions, especially in pressure prediction, where the mean absolute percentage error (MAPE) is reduced by 87.8% and 92.7% respectively.

A Knowledge-Inspired Hierarchical Physics-Informed Neural Network for Pipeline Hydraulic Transient Simulation

Physics-informed neural networks for hydraulic transient analysis in pipeline systems

Reconstructing transient pressures in pipe networks from local observations by using physics-informed neural networks

Deep Learning Based Hydraulic-mechanical Simulation of Petroleum Reservoirs

Data-driven Computation of Natural Gas Pipeline Network Hydraulics

Surrogate Modeling for Porous Flow Using Deep Neural Networks

A Novel Sequential-Based Hybrid Approach Incorporating Physical Modeling and Deep Learning for Multiphase Subsurface Flow Simulation

Deeppipe: Theory-guided LSTM method for monitoring pressure after multi-product pipeline shutdown

Deeppipe: A Two-Stage Physics-Informed Neural Network for Predicting Mixed Oil Concentration Distribution

Deeppipe: Theory-guided neural network method for predicting burst pressure of corroded pipelines

Efficient Super-Resolution of Pipeline Transient Process Modeling Using the Fourier Neural Operator

A Temporal and Spatial Prediction Method for Urban Pipeline Network Based on Deep Learning

Impact of Uncertainty in the Physics-Informed Neural Network on Pressure Prediction for Water Hammer in Pressurized Pipelines

A Physics-Informed Spatial-Temporal Neural Network for Reservoir Simulation and Uncertainty Quantification

Numerical model for calculation of hydraulic transients with two-phase flow and fluid–structure interaction

A Novel Reservoir Simulation Model Based on Physics Informed Neural Networks

Physics Informed Deep Learning for Flow and Transport in Porous Media

Integrated Design and Operational Optimization of Gathering Pipeline Networks Using Neural Network Surrogate Models

A Gradient-based Deep Neural Network Model for Simulating Multiphase Flow in Porous Media

An innovative end-to-end PINN-based solution for rapidly simulating homogeneous heat flow problems: An adaptive universal physics-guided auto-solver

Multiphase flow prediction with deep neural networks