Codes de calcul (structure) explicite
Théorie
Méthode de résolution des problèmes transitoires "rapides". Le schéma est conditionnellement stable c'est à dire que la stabilité est assurée si et seulement si le pas de temps est inférieur à un pas dit de "stabilité". Ce pas de temps correspond au temps de passage d'une onde au travers d'un élément fini. Il existe des méthodes pour augmenter ce pas de stabilité (mass scaling). Néanmoins le pas de temps dépend de nombreux autres paramètres tels que, par exemple le chargement.
Ce pas de temps est en général très petit (taille d'un élément fini divisé par vitesse des ondes), d'où pas besoin d'itérations pour assurer l'équilibre de chaque pas de temps.
En "échange", la solution ne dépend pas, dans le cas général (pas d'amortissement, multiplicateurs de Lagrange) de l'inversion d'un système linéaire. Le calcul des matrices élémentaires de rigidité est néanmoins nécessaire.mais réduit car, pour d'autres raisons, les éléments sont, souvent, sous-intégrés.
Bibliographie
- An explicit finite element primer (NAFEMS, 2002) - en anglais
- Paul Jacob, Lee Goulding.
- Explicit Finite Element Methods for Large Deformation Problems in Solid Mechanics (Wiley, 2007) - en anglais
- Introduction to the Explicit Finite Element Method for Nonlinear Transient (Wiley, 2012) - en anglais
- Shen R. Wu, Lei Gu
Benchmarks
-
Explicit Finite Element Code Verification Problems (Miles Buechler, Amanda McCarty, Derek Reding, Ryan Maupin)
-
–1948 - The use of flat-ended projectiles for determining dynamic yield stress I. Theoritical considerations – (Geoffrey Ingram Taylor) Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences, 194, p. 289-299–1948 - … II. Tests on various metallic materials (A. C. Whiffin)–1948 - … III. Changes in microstructure caused by deformation under impact at high-striking velocities (W. E. Carrington, Marie L. V. Gayler)
Codes
En 1987, Lee M. Taylor et Dennis P. Flanagan (co-développeurs de PRONTO au laboratoire de Sandia) développent la première version de ABAQUS-Explicit.
In 1983, the advent of the IBM PC provided an opportunity for Birnbaum and Cowler
to use their combined expertise in the development of what would later become the
first fully integrated (pre, post and solution) and interactive explicit coupled
nonlinear dynamics code. With this goal in mind, they left PISCES International and
in 1985 co-founded Century Dynamics Inc., with AUTODYN version 1.0 as the first
hydrocode developed on a PC platform and a few years later available on larger
computing systems.On January 5, 2005, ANSYS, Inc. announced the acquisition of Century Dynamics
Inc., a software company headquartered in Concord, California. Century Dynamics
creates software for the explicit dynamics and oil and gas markets. This article
explores the history of the company, explains where the people and the software
came from, and presents an overview of the AUTODYN explicit dynamics product.
Code développé au LLL par John O. Hallquist dans les années 76. 1978 domaine public. David Benson à partir de 1984. 19888 LS Dyna par LSTC (Livermore Software Technology Corporation) https://www.d3view.com/wp-content/uploads/2007/06/benson.pdf
Code développé depuis par le CEA sous l’impulsion d’Alain Hoffmann qui confie la mission à Michel Lepareux (parti en retraite en 2008). Comme les codes développés au CEA, la commercialisation est confiée à CISI Ingénierie puis à SOCOTEC-Industrie.
Parallèlement le CCR d’ISPRA développe PLEXIS (sur base PLEXUS orienté Fluide-Structure suite à des accords initiés par Alain Hoffmann) sous la direction de Jean-Pierre Halleux (parti en retraite en 2008) et Folco Casadeï. En 1999, Michel Géradin (devenu Directeur du laboratoire ELSA du CCR, parti en retraite en 2010) encourage la création du consortium EuroPlexus : SAMTECH devient le distributeur d’EuroPlexus. Longtemps utilisé par la SNECMA (années 90 sous l’impulsion de Jean-Pierre Mascarell) qui a des accords directs avec le CEA et recrute des thésards tels que : Jerôme Bonini ou Stéphane Giusti ou des spécialistes comme Eric Seinturier (ex SOCOTEC-Industries qui part en 2004 chez Turboméca à la Direction Technique), EuroPlexus est aujourd’hui très utilisé par EDF R&D (Serguei Potapov) et SEPTEN (Yannick Pons après Emmanuel Viallet et Claude Duval)
Acronyme de Hydrodynamics Elastic Magneto Plastic. Code développé au LLL par Mark L. Wilkins dans les années 70. Auteur en 1999 chez SPRINGER de Computer simulation of dynamic phenomena
The HEMP 3D program can be used to solve problems in solid mechanics involving dynamic plasticity and time dependent material behavior and problems in gas dynamics. The equations of motion, the conservation equations, and the constitutive relations listed below are solved by finite difference methods following the format of the HEMP computer simulation program formulated in two space dimensions and time
Code développé sur les bases de DYNA 3D par John O. Hallquist en 1987 quand il crée Livermore Software Technology Corporation (LSTC) http://www.dynasupport.com/manuals. https://www.d3view.com/wp-content/uploads/2007/06/benson.pdf
Code développé par le CEA-DAM (Direction des Applications Militaires)
PAM-Crash a été développé dans les années 80 par Eberhard Haug et son équipe sur la base du programme DYNA3D (dans le domaine public). Lire la suite…
As is the case for many of the leading companies in today’s CAE market, Century
Dynamics’ history can be traced to the defense and nuclear industries. During the
1950s and 1960s, finite difference physicists in the United States developed the core
theories for what would later be the foundation of explicit dynamics software
programs. The first of these explicit dynamics programs were known as
“hydrocodes,” technology then used exclusively at government laboratories, such as
Lawrence Livermore and Los Alamos, to model events of interest to the defense
community.
Physics International, founded in 1963 by scientists who left Lawrence Livermore
Laboratory, developed in-house hydrocodes for their own defenserelated consulting.
Recognizing the commercial need for the technology in nuclear safety applications,
the PISCES International division was set up in 1971, with young physicist Naury
Birnbaum as its president. At about the same time, the United Kingdom Atomic
Energy Authority (UKAEA) funded development of its own hydrocodes that could be
used to investigate the potential consequence of core disruptive accidents in nuclear
power plants. Malcolm Cowler, a key developer of the UK codes, joined the PISCES
International team in 1977. The main product of PISCES International was the
coupled Euler- Lagrange explicit finite difference code PISCES, which at the time was
the only commercially available computer code that could solve detailed fluid
structure interaction problems.
In 1983, the advent of the IBM PC provided an opportunity
Code éléments finis développé au Laboratoire de Sandia jusqu’en 1991 par Lee M. Taylor et Dennis P. Flanagan puis repris par Steve Plimpton.
PRONTO is a three-dimensional, transient solid dynamics code which is used for analyzing large deformations of nonlinear materials subjected to high rates of strain[3]. Developed over the past 10 years, PRONTO is a production-level code used by over 50 organizations inside and outside Sandia. Input to the code includes an unstructured grid consisting of an arbitrary mixture of hexahedral elements, shell elements, rigid bodies and smoothed particles. PRONTO implements a Lagrangian finite-element method with explicit time integration and adaptive timestep control to integrate the equations of motion. The finite-element formulation uses eight-node, uniform strain hexahedral elements and four-node quadrilateral uniform strain shell elements. Either the Flanagan-Belytschko hourglass control scheme or an assumed-strain hourglass control scheme can be used to control element distortions. PRONTO contains a variety of complex, nonlinear material models, including elastic-plastic materials with various types of strain hardening. A critical feature of the code is a robust algorithm for detecting when one material surface contacts another, for example in an automobile collision when the bumper buckles into the radiator. Correctly identifying surfaces in contact requires sophisticated algorithms for searching the global set of finite-elements. In a complex simulation, the cost of contact detection alone can be more than 50% of the run time on a sequential machine. A PRONTO timestep has the following structure.
1. Perform finite element analysis to compute forces on elements.
2. Compute forces between smoothed particles.
3. Predict new locations of particles and grid elements.
4. Search for contacts between mesh elements, or between elements and particles.
5. Correct the locations by pushing back objects in contact.
Stages (1), (2) and (4) dominate the sequential run time. The contact search in stage (4) typically consumes 30-60% of the time, so a great deal of effort has been expended over the years to make this computation fast[4]. The result of this effort was the replacement in PRONTO of floating point operations with a faster approach involving sorting and searching in integer lists.
Code développé par la société MECALOG, sous l’impulsion de Francis Arneaudeau (ex ESI)