ROSE-pyforce

A Python Framework for Data-Driven Model Order Reduction in Multi-Physics Problems

ROSE-pyforce is an open-source Python framework designed for Data-Driven Reduced Order Modeling (DDROM), particularly in the context of multi-physics problems in nuclear engineering. Built upon the FEniCSx project, ROSE-pyforce utilizes the dolfinx package for key computational tasks, including mesh generation, integral calculations, and function storage (Riva et al., 2024).

🔑 Key Features

🚀 Data-Driven Reduced Order Modeling (DDROM)

ROSE-pyforce reduces the computational complexity of multi-physics simulations by integrating real measurement data, improving the accuracy and efficiency of surrogate models (Riva et al., 2024).

Illustration of Data-Driven Reduced Order Modeling methodologies (Riva et al., 2024).

🎯 Sensor Placement Optimization

ROSE-pyforce includes algorithms that optimize sensor placement, crucial for accurate data assimilation and model calibration in complex physical systems (Cammi et al., 2024).

🔓 Open-Source Accessibility

The framework is openly available on GitHub, encouraging collaboration and further development in the scientific community.

ROSE-pyforce represents a significant advancement in the use of data-driven techniques for model order reduction, offering an innovative and computationally efficient approach for multi-physics problems. 🚀

References

2024

JOSS

pyforce: Python Framework for data-driven model Order Reduction of multi-physiCs problEms

Stefano Riva, Carolina Introini, and Antonio Cammi

under review at Journal of Open Source Software, Jul 2024

DOI Bib

@article{pyforce2024,
  doi = {},
  url = {},
  year = {2024},
  publisher = {The Open Journal},
  volume = {},
  number = {},
  month = jul,
  pages = {},
  author = {Riva, Stefano and Introini, Carolina and Cammi, Antonio},
  title = {pyforce: Python Framework for data-driven model Order Reduction of multi-physiCs problEms},
  journal = {under review at Journal of Open Source Software},
}

AMM
Multi-physics model bias correction with data-driven reduced order techniques: Application to nuclear case studies

Stefano Riva, Carolina Introini, and Antonio Cammi

Applied Mathematical Modelling, Nov 2024

Abs DOI Bib

Due to the multiple physics involved and their mutual and complex interactions, nuclear engineers and researchers are constantly working on developing highly accurate Multi-Physics models, focusing in particular on the core coupling of Neutronics and Thermal-Hydraulics. Nevertheless, the development of accurate and stable models remains a challenging task despite the advancements in computational hardware and software. This work investigates the possibility of combining the available mathematical model with data collected on physical systems, with a two-fold goal: improvement of the performance of the former from the computational point of view without sacrificing accuracy and performing model bias correction with the knowledge coming in real-time from the system. In particular, two Data-Driven Reduced Order Modelling techniques, the Generalised Empirical Interpolation Method and the Parametrised-Background Data-Weak formulation, are applied to literature benchmark nuclear case studies, as they were observed to be quite well suited for the chosen cases. The main goal of this work is to assess the possibility of using external data to perform model bias correction: starting from a purposefully less accurate, but computationally cheaper base model, then using high-fidelity data to update and correct the model efficiently. Indeed, the numerical results obtained in this paper are promising, confirming the feasibility of this approach to develop computationally cheap and accurate multi-physics models; furthermore, investigation of Data-Driven Reduced Order Modelling approach to nuclear industrial cases, in the context of model bias correction, is foreseen.
@article{RIVA2024243, title = {Multi-physics model bias correction with data-driven reduced order techniques: Application to nuclear case studies}, journal = {Applied Mathematical Modelling}, volume = {135}, pages = {243-268}, year = {2024}, issn = {0307-904X}, doi = {https://doi.org/10.1016/j.apm.2024.06.040}, month = nov, url = {https://www.sciencedirect.com/science/article/pii/S0307904X24003196}, author = {Riva, Stefano and Introini, Carolina and Cammi, Antonio}, keywords = {Reduced order modelling, Data driven, Nuclear reactors, Multi-physics, Model correction} }
NED
Data-driven model order reduction for sensor positioning and indirect reconstruction with noisy data: Application to a Circulating Fuel Reactor

Antonio Cammi, Stefano Riva, Carolina Introini, and 2 more authors

Nuclear Engineering and Design, May 2024

Abs DOI Bib

Sensor positioning and real-time estimation of non-observable fields is an open question in the nuclear sector, especially for advanced nuclear reactors. In Circulating Fuel Reactors (CFR), liquid fuel and coolant are homogeneously mixed, and thus these reactors will not have internal structures, making sensor positioning in the primary circuit, including the core, an unresolved problem, making most of the core blind to sensors. Thus, the possibility of estimating the system state in the whole domain using a few local measurements has important implications for safety, monitoring, and control both in nominal and accidental conditions. In this context, the integrated Model Order Reduction and Data Assimilation framework offers intriguing opportunities to reliably combine experimental data and background knowledge from a reduced mathematical model. This work discusses and applies innovative methods within this framework, based on the Generalized Empirical Interpolation and the Indirect Reconstruction algorithms, to a proposed concept of CFR. This work aims to identify the optimal sensor positioning within the core and assess the feasibility of reconstructing the quantities of interest starting only from transient sparse data on fuel temperature, possibly noisy, and testing the predictive capabilities of the discussed methods.
@article{DDMOR_CFR, title = {{Data-driven model order reduction for sensor positioning and indirect reconstruction with noisy data: Application to a Circulating Fuel Reactor}}, journal = {Nuclear Engineering and Design}, volume = {421}, pages = {113105}, year = {2024}, issn = {0029-5493}, doi = {https://doi.org/10.1016/j.nucengdes.2024.113105}, month = may, url = {https://www.sciencedirect.com/science/article/pii/S002954932400205X}, author = {Cammi, Antonio and Riva, Stefano and Introini, Carolina and Loi, Lorenzo and Padovani, Enrico}, keywords = {Hybrid Data-Assimilation, Generalized Empirical Interpolation Method, Indirect Reconstruction, Sensors positioning, Molten Salt Fast Reactor, Noisy data} }