Physics-Informed Neural Network for Reconstructing Free-Surface Transient Flow Fields in Long-Distance Water-Conveyance Tunnels from Sparse Observations
Физически информированная нейронная сеть для реконструкции нестационарных полей течения со свободной поверхностью в протяжённых водопроводящих тоннелях по разреженным наблюдениям
2026-09-07
SCID: 54.1/vkzt2f7u
Discuss with AI
Darcy–Weisbach frictionSaint-Venant equationsfree-surface transient flowphysics-informed neural network (PINN)sparse sensor reconstruction
Figures from the paper
Abstract (AI)
Long-distance free-surface water-conveyance tunnels require a spatially continuous representation of transient water depth, yet in practice flow is monitored at only a few sections. This study develops a physics-informed neural network (PINN) that reconstructs the transient water-depth field of unsteady free-surface flow in such tunnels from two- or three-point sensors. The one-dimensional Saint-Venant equations, closed with a Darcy–Weisbach steady friction term in hydraulic-radius form, are embedded as a soft constraint in the training loss, so that sparse depth observations are combined with the governing conservation laws to recover the field at unobserved interior and downstream locations. High-resolution finite-volume (FVM) solutions of a 500 m circular tunnel under a flood-rise scenario provide the reference data. The PINN reduces the relative L2 error at unobserved sections to approximately one-quarter of that of an otherwise identical, physics-free ANN (2.50% versus 10.05% at an interior section; 4.77% versus 13.69% at an extrapolation section). Runs repeated with different random seeds confirm statistical stability at zero noise, while revealing that a minority of trainings at 10% noise converge to spurious solutions. Sensor-placement experiments, including layouts anchored at the true domain boundaries (x = 0 and 500 m), show that boundary anchoring—particularly of the upstream boundary—governs both accuracy and noise robustness: boundary-anchored two-sensor layouts remain accurate in most runs under 10–20% observation noise, whereas interior-only layouts degrade sharply. The method is presented as an offline reconstruction tool; its extension to streaming data assimilation is discussed as future work.
Key Findings
1
A physics-informed neural network reconstructs transient water-depth fields in long-distance free-surface tunnels using only two or three sparse sensor locations.
2
At 10% observation noise, most boundary-anchored two-sensor runs remain accurate, but a minority of trainings converge to spurious solutions; streaming data assimilation remains future work.
3
Boundary-anchored sensor layouts, especially those including the upstream boundary, provide greater accuracy and noise robustness than interior-only layouts.
4
Embedding the one-dimensional Saint-Venant equations with Darcy–Weisbach friction substantially improves reconstruction over an otherwise identical physics-free ANN.
5
Relative L2 errors decrease from 10.05% to 2.50% at an interior section and from 13.69% to 4.77% at an extrapolation section.
Research Object
transient free-surface flow in a long-distance water-conveyance tunnel
Research Subject
spatial reconstruction accuracy and noise robustness of the transient water-depth field from sparse observations, governed by sensor placement and physics-informed constraints
Publication Details
Publication Date
2026-09-07
Journal
Publisher
ISSN
Cited by
0
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai3
Understanding and Mitigating Gradient Flow Pathologies in Physics-Informed Neural Networks2021
Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s Next2022
Physics‐Informed Neural Networks of the Saint‐Venant Equations for Downscaling a Large‐Scale River Model2023