 |
|||
| Url | https://cimne.com/sgp/rtd/Project.aspx?id=921 | ||
|
|||
| Acronym | GRAIN | ||
| Project title | An innovative multi-scale data-driven paradigm for the modelling of granular flows | ||
| Reference | PID2021-122676NB-I00 | ||
| Principal investigator |
Alessandro FRANCI - falessandro@cimne.upc.edu
Juan Marcelo GIMENEZ - jmgimenez@cimne.upc.edu |
||
| Start date | 01/09/2022 | End date | 31/08/2025 |
| Coordinator | CIMNE | ||
| Consortium members |
|
||
| Program | P.E. para Impulsar la Investigación Científico-Técnica y su Transferencia | Call | Proyectos Generación de Conocimiento 2021 |
| Subprogram | Subprograma Estatal de Generación de Conocimiento | Category | Nacional |
| Funding body(ies) | MICINN | Grant | $100,430.00 |
| Abstract | Granular materials are ubiquitous in nature and are the most handled material in the industry after water. Their particulate nature gives them unique physical properties and makes the modelling of their behaviour a challenging task. Given the difficulty of approaching granular media from the point of view of their individual particles, the current modelling tools tend to consider granular materials as a continuum. These methods use general constitutive laws, which are unable to capture those grain-scale phenomena that drive their complex and ambiguous behaviour. The continuous models, straying from the real nature of granular media, are suitable only for limited motion regimes and cannot be used for detailed engineering studies. On the other hand, when discrete numerical models are used, simplifying hypotheses are commonly assumed to overcome the otherwise unaffordable computational costs, such as considering grains of spherical shapes and using coarse-graining techniques. These abridged methods, besides requiring complex calibrations, sacrifice the distinctive capability of discrete numerical methods of faithfully reproducing the particle-particle interactions, in exchange for a reduced computational cost of the analyses. In conclusion, nowadays there is still the tendency of distorting the available numerical methods in order to make them applicable, although with significant limitations, to granular media modelling. GRAIN aims to reverse this perspective. In GRAIN, all the numerical methods are used within their comfort zone and are put at service of the granular material essence. This challenging objective will be achieved through a wise combination of innovative computational methods and machine learning toolsthat will allow modelling the granular material starting from the particle-particle interactions. In particular, high-fidelity microscale computations on granular assemblies will be realised with a discrete numerical method to capture the grain-scale phenomena. In these representative volume elements, the real shape and size of the grains will be considered. The material response will then be converted into a dataset which will be used to train an artificial neural network. This surrogated model will provide the state of stress of the granular media from a state of deformation. Due to the importance of granular materials in thermal-coupled applications, the same scheme will be used to predict the heat flux through the granular assembly from a given temperature field. The microscale information will be used by the macroscale modelling tool based on a continuum numerical method to predict the response of the whole granular media. This way, without using any constitutive law, this multi-scale data-driven model will be able to reproduce the grain-scale phenomena also in large-scale problems. Thanks to its generality, the GRAIN paradigm can be applied to granular media of different types, from powders to rock avalanches, and therefore to varied industrial and engineering applications. Based on their widespread presence in nature and the industry, GRAIN will focus on dense granular flows. From a long-term perspective, this new granular media model will enable the optimisation of varied industrial processes and products, the design of new materials with enhanced properties, and the realisation of accurate geotechnical engineering studies, leading to tangible benefits in terms of costs, material and energy consumptions and safety | ||
| Ayuda PID2021-122676NB-I00 del proyecto financiado por MCIN/AEI/10.13039/501100011033/ y por "FEDER: Una manera de hacer Europa" | |||