Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2509.01189 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914015435816960 |
|---|---|
| author | Wang, Xicheng Chan, YiMeng Wong, KinWing Grishchenko, Dmitry Kudinov, Pavel |
| author_facet | Wang, Xicheng Chan, YiMeng Wong, KinWing Grishchenko, Dmitry Kudinov, Pavel |
| contents | Measurement of the velocity field in thermal-hydraulic experiments is of great importance for phenomena interpretation and code validation. Direct measurement employing Particle Image Velocimetry (PIV) is challenging in some multiphase scenarios where the measurement system would be strongly affected by the phase interaction. In such cases, measurement can only be achieved via sparsely distributed sensors, such as Thermocouples (TCs) and pressure transducers. An example can refer to steam injection into a water pool where the rapid collapse of bubbles and significant temperature gradient make it impossible to obtain the main flow velocity at a large steam flux by PIV. This work investigates the feasibility and capability of utilization of data-driven modeling for flow reconstruction from sparse temperature data. The framework applies (i) a Proper Orthogonal Decomposition (POD) to encode variables from full space to latent space and (ii) a Fully connected Neural Network (FNN) to approximate sparse measurements to coefficients of latent space. Sensor positioning aiming to identify the optimal sensor location is also discussed. The proposed framework has been tested on a single-phase planar jet and steam condensing jets issued through a multi-hole sparger. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_01189 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Data-driven modeling for flow reconstruction from sparse temperature measurements Wang, Xicheng Chan, YiMeng Wong, KinWing Grishchenko, Dmitry Kudinov, Pavel Fluid Dynamics Measurement of the velocity field in thermal-hydraulic experiments is of great importance for phenomena interpretation and code validation. Direct measurement employing Particle Image Velocimetry (PIV) is challenging in some multiphase scenarios where the measurement system would be strongly affected by the phase interaction. In such cases, measurement can only be achieved via sparsely distributed sensors, such as Thermocouples (TCs) and pressure transducers. An example can refer to steam injection into a water pool where the rapid collapse of bubbles and significant temperature gradient make it impossible to obtain the main flow velocity at a large steam flux by PIV. This work investigates the feasibility and capability of utilization of data-driven modeling for flow reconstruction from sparse temperature data. The framework applies (i) a Proper Orthogonal Decomposition (POD) to encode variables from full space to latent space and (ii) a Fully connected Neural Network (FNN) to approximate sparse measurements to coefficients of latent space. Sensor positioning aiming to identify the optimal sensor location is also discussed. The proposed framework has been tested on a single-phase planar jet and steam condensing jets issued through a multi-hole sparger. |
| title | Data-driven modeling for flow reconstruction from sparse temperature measurements |
| topic | Fluid Dynamics |
| url | https://arxiv.org/abs/2509.01189 |