Journal article 85 views
A novel finite-element-based solution and property construction method for thermal problems from sparse data
Wiera Bielajewa,
Michelle Baxter,
Perumal Nithiarasu 
International Journal of Numerical Methods for Heat and Fluid Flow
Swansea University Authors:
Wiera Bielajewa, Perumal Nithiarasu
Abstract
Sparse experimental measurements from diagnostic sensors are often the only source of data available during an experiment. To enable monitoring and control of such experiments (digital twinning) rapidly estimating the full field solution and material properties using sparse data may be useful, espec...
| Published in: | International Journal of Numerical Methods for Heat and Fluid Flow |
|---|---|
| Published: |
|
| URI: | https://cronfa.swan.ac.uk/Record/cronfa70855 |
| Abstract: |
Sparse experimental measurements from diagnostic sensors are often the only source of data available during an experiment. To enable monitoring and control of such experiments (digital twinning) rapidly estimating the full field solution and material properties using sparse data may be useful, especiallyunder extreme thermal environments. This paper addresses such a construction procedure using an efficient finite-element-based approach combined with a modified ODIL (Optimizing a DIscrete Loss) concept. A finite element specific regularisation term is added to the loss function to resolve the ill-posedness. The loss function gradients are calculated analytically. The nonlinear material properties are constructed as a piecewise linear function during the temperature change. A sample from fusion energy experimental facility is used as the test case to demonstrate the proposed methodology. The results indicate that near real-time solution construction is possible, which makes this approach suitable for digital twinning. |
|---|---|
| College: |
Faculty of Science and Engineering |
| Funders: |
EPSRC Energy Programme [grant number EP/W006839/1].
The authors acknowledge the support of Supercomputing Wales and AccelerateAI projects, which is part-funded by the European Regional Development Fund (ERDF) via the Welsh Government.
Furthermore, the authors gratefully acknowledge NVIDIA Academic Grant Program Award for supporting this research through the NVIDIA RTX 6000 Ada GPU grant. |