Summary
The fundamental importance of materials to modern society is evidenced by the way new materials have revolutionized almost every aspect of our lives. Despite the many advances, dwindling resources and more stringent demands on product cost and performance demand increasingly better material designs and production processes, resulting in a heightened reliance on computational methods.
In the field of computational materials engineering, the recent emergence of data science into the mainstream is causing a paradigm shift in the way models and data are used. There is a shift from traditional simulation methods which use data mainly to calibrate parameters in models, to data-driven simulation methods which seek to bypass the use of models by extracting knowledge from large data sets. This project synergistically combines aspects of both – by developing advanced computational methods that permit multi-scale material models to be informed by available measurement data.
This project addresses this challenging problem through two main tasks. In the first part, we develop dimension reduction techniques for rapid multi-scale materials simulations. These methods must be capable of dealing with deterministic and stochastic microstructure parameters reflecting variations in loading, material, and morphological properties. In the second part, the reduced order models serve as an enabler for the development of computational methods for the selection of the most informative data and its assimilation into multi-scale material models. By enabling parameter estimation and model correction, this leads to increased accuracy and precision in the prediction of engineering quantities of interest.
The success of the project will give rise to a novel computational framework that enables real-time multi-scale materials simulations informed by optimally chosen data, thus permitting effective risk management and cost reduction in the design of materials and control of manufacturing processes.
In the field of computational materials engineering, the recent emergence of data science into the mainstream is causing a paradigm shift in the way models and data are used. There is a shift from traditional simulation methods which use data mainly to calibrate parameters in models, to data-driven simulation methods which seek to bypass the use of models by extracting knowledge from large data sets. This project synergistically combines aspects of both – by developing advanced computational methods that permit multi-scale material models to be informed by available measurement data.
This project addresses this challenging problem through two main tasks. In the first part, we develop dimension reduction techniques for rapid multi-scale materials simulations. These methods must be capable of dealing with deterministic and stochastic microstructure parameters reflecting variations in loading, material, and morphological properties. In the second part, the reduced order models serve as an enabler for the development of computational methods for the selection of the most informative data and its assimilation into multi-scale material models. By enabling parameter estimation and model correction, this leads to increased accuracy and precision in the prediction of engineering quantities of interest.
The success of the project will give rise to a novel computational framework that enables real-time multi-scale materials simulations informed by optimally chosen data, thus permitting effective risk management and cost reduction in the design of materials and control of manufacturing processes.
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/818473 |
| Start date: | 01-01-2020 |
| End date: | 30-06-2026 |
| Total budget - Public funding: | 1 999 632,00 Euro - 1 999 632,00 Euro |
Cordis data
Original description
The fundamental importance of materials to modern society is evidenced by the way new materials have revolutionized almost every aspect of our lives. Despite the many advances, dwindling resources and more stringent demands on product cost and performance demand increasingly better material designs and production processes, resulting in a heightened reliance on computational methods.In the field of computational materials engineering, the recent emergence of data science into the mainstream is causing a paradigm shift in the way models and data are used. There is a shift from traditional simulation methods which use data mainly to calibrate parameters in models, to data-driven simulation methods which seek to bypass the use of models by extracting knowledge from large data sets. This project synergistically combines aspects of both – by developing advanced computational methods that permit multi-scale material models to be informed by available measurement data.
This project addresses this challenging problem through two main tasks. In the first part, we develop dimension reduction techniques for rapid multi-scale materials simulations. These methods must be capable of dealing with deterministic and stochastic microstructure parameters reflecting variations in loading, material, and morphological properties. In the second part, the reduced order models serve as an enabler for the development of computational methods for the selection of the most informative data and its assimilation into multi-scale material models. By enabling parameter estimation and model correction, this leads to increased accuracy and precision in the prediction of engineering quantities of interest.
The success of the project will give rise to a novel computational framework that enables real-time multi-scale materials simulations informed by optimally chosen data, thus permitting effective risk management and cost reduction in the design of materials and control of manufacturing processes.
Status
SIGNEDCall topic
ERC-2018-COGUpdate Date
27-04-2024
Images
No images available.
Geographical location(s)
