DemoSim project

Democratizing numerical simulation in engineering and applied sciences

Project DPI2015-72365-EXP

MEC

Participants:

Group of Applied Mechanics and Bioengineering. amb-I3A.

Elias Cueto

Iciar Alfaro

David Gonzalez

Francisco Chinesta (GeM EC-Nantes)

Goals of the project:

The complexity of nowadays numerical simulation in industry and aplied science, together with the increasing importance of big data, coming from both experiments (online measurements) and simulations, in every aspect of research is giving rise to the appearance of a sort of “custodians” or gurus of the principles and know-how of numerical simulation. Industry, on the contrary, is increasingly interested on democratizing the practice and analysis of simulation within their design cycle. Thus, a faster and easier-to-analyze simulation procedure is actively sought. Model order reduction has been able to obtain impressive savings in computer time. However, the vast amount of simulation data (along with online measurements) we are producing, makes this reduction unpractical if we do not develop fast ways to extract relevant information from this amount of (big) data. DemoSIM aims at developing unprecedented numerical techniques that both reduce the complexity of today’s CAE, therefore democratizing their use all along the product development team, with a minimum amount of computer cost and, more importantly, to capture and re-use knowledge, the relevant information, in order to be incorporated in an authomatized way to the industry know how. This will be achieved by developing a new family of techniques involving both model reduction techniques and machine learning strategies, so as to be able to extract the manifold structure revealing the (small) relevant information coming from all this big data flow. These unprecedented techniques must be able to go beyond current model order reduction techniques, that have revealed very important improvements in the simulation pipeline, but have also revealed very important deficiencies in some applications.

Core objectives:

  1. Development of a posteriori NLDR simulation techniques
  2. Development of a priori NLDR techniques
  3. Application in surgery planning
  4. Image-based analysis of micro and nanostructured materials

Refereed Journal Publications

FORTHCOMING

  1. Reduced order modeling for physically-based augmented reality. A. Badías, I. Alfaro, D. Gonzalez, F. Chinesta, E. Cueto. Computer Methods in Applied Mechanics and Engineering, in press, 2018. [Download pdf of draft] [video1][video2][video3]

2018

  1. Haptic simulation of tissue tearing during surgery. C. Quesada, I. Alfaro, D. Gonzalez, F. Chinesta, E. Cueto. International Journal for Numerical Methods in Biomedical Engineering, 34 (3), e2926. [Download pdf of draft][video1][video2]
  2. Reduced-order modeling of soft robots. Jean Chenevier, David Gonzalez, Jose Vicente Aguado, Francisco Chinesta and Elias Cueto. PLoS ONE, 13(2): e0192052, 2018. [Download PDF of draft] [OpenAccess]
  3. A Manifold Learning Approach to Data-Driven Computational Elasticity and Inelasticity. R. Ibañez, E. Abisset-Chavanne, J. V. Aguado, D. Gonzalez, E. Cueto, F. Chinesta. Archives of Computational Methods in Engineering, 25(1), 47-57, 2018. [Download pdf of draft].
  4. A manifold learning approach for Integrated Computational Materials Engineering. E. Lopez, D. Gonzalez, J.V. Aguado, E. Abisset-Chavanne, F. Lebel, R. Upadhyay, E. Cueto, C. Binetruy, F. Chinesta. Archives of Computational Methods in Engineering, 25(1), 59-68, 2018. [Download pdf of draft]
  5. kPCA-based Parametric Solutions within the PGD Framework. D. Gonzalez, J.V. Aguado, E. Cueto, E. Abisset-Chavanne, F. Chinesta. Archives of Computational Methods in Engineering, 25(1), 69-86, 2018. [Download pdf of draft]

2017

  1. Local Proper Generalized Decomposition. A. Badías, D. González, I. Alfaro, F. Chinesta, E. Cueto. International Journal for Numerical Methods in Engineering, 112:12,1715–1732, 2017. [Download pdf of draft]
  2. Model order reduction for real-time data assimilation through Extended Kalman Filters. D. Gonzalez, A. Badias, I. Alfaro, F. Chinesta, E. Cueto. Computer Methods in Applied Mechanics and Engineering, 326, 679-693, 2017. [Download pdf of draft]
  3. Data-driven non-linear elasticity. Constitutive manifold construction and problem discretization. R. Ibañez, D. Borzacchiello, J. V. Aguado, E. Abisset-Chavanne, E. Cueto, P. Ladeveze, F. Chinesta. Computational Mechanics, 60 (5), 813–826, 2017. [Download pdf of draft]

2016

  1. Computational patient avatars for surgery planning. D. Gonzalez, E. Cueto and F. Chinesta. Annals of Biomedical Engineering, 44(1), 35-45. 2016. [Download pdf of draft]

Our work in the media

Data-enabled, Physics-constrained Predictive Modeling of Complex Systems. SIAM News, July 2017. [link]

Proudly edited the special issue of Archives of Computational Methods in Engineering (Springer) on Machine Learning in Computational Mechanics. See the issue here.