Salome7 the Open Source Integration Platform for Numerical Simulation Salome7 Platform

salome-platform.org

SALOME7 THE OPEN SOURCE INTEGRATION PLATFORM FOR NUMERICAL SIMULATION SALOME7 PLATFORM

The SALOME software platform is an open framework that can integrate scientific solvers for modelling various physical domains and realize a wide range of computing simulations. In the context of CEA and EDF, the physical systems are typically the industrial equipments of the nuclear production plants, with as primary concern the design of new generation reactors, the nuclear fuel management, the reliability and safety of the installations, or the equipments ageing for an optimized life cycle management. The challenges are to analyse the behaviour of the systems with sufficient accuracy and performance to aid the decision making on industrial questions. SALOME comes natively with CAD modelling, mesh algorithms, 3D visualization and advanced functions for simulation data processing.

Figure 1: Screenshot of the public web site of SALOME. www.salome-platform.org KEY FEATURES MODEL OF DOWNLOAD THE The numerical analysis of complex DEVELOPMENT SALOME PLATFORM industrial systems demands from The SALOME platform is actively The SALOME platform can be scientists and engineers to inte- developed by CEA and EDF, two of downloaded from the web site: grate most domains of physics like the key actors in the domain of http://www.salome-platform.org material science, solid mechanics energy in France, and with the for several LINUX distributions and and structural dynamics, fluids support of the EURIWARE/Open WINDOWS. The site provides tuto- physics and thermohydraulics, nu- Cascade company, a leader in the rials, a forum section and gives ac- clear physics and radiations, elec- development of software programs cess to user documentation and tromagnetism. These domains can for scientific modelling and com- source code under LGPL licence. be federated in one single simula- puting simulation. This 12 years old tion environment, the SALOME partnership provides the SALOME SERVICE AND SUPPORT platform. development project with a very EURIWARE and Open Cascade pro- The main features of the SALOME committed and dedicated team in vide a whole range of services for platform are : scientific computing, in particular SALOME towards professional end- for specific applications in nuclear users including technical support The design of the geometry industry. and specific training. representation of physical systems (CAD modelling) and its as- Moreover, the SALOME platform Support services are available sociated discretized model development relies on the integra- within a “à-la-carte” support pro- (meshing functions for finite ele- tion of cutting edge third party gram particularly suited for univer- ments or finite volumes solvers). software programs, for example sities and academic organizations the commercial advanced and ro- as well as for small or larger indus- The ability to integrate do- DEVELOPMENT bust meshing programs Mesh- trial companies: main specific solvers into Gems-CADSurf (formerly known as FACT SHEET normalized software compo- Helpdesk support for expert BLSURF) and MeshGems- Tetra (for- nents with standard interfaces to needs concerning a one-shot Project kick off 2001 merly known as GHS3D) (by the facilitate the coupling of different technical issue, delivered by mail Distene Company) and the PAR- Development 20 persons physical domains. or by phone within a guaranteed team AVIEW software including the VTK time frame. The supervision of compu- 3D visualisation toolkit (by the Technical support for complex Licence LGPL tation workflows defined as Kitware Company). graphs of distributed software problem solving that requires the Recent efforts in the development Distribution one major components, including CAD help of a qualified engineer. of SALOME for parallel computa- version every modelling, domain specific Expert consulting delivered on tion have been supported by 2 years, and solvers and data processing com- the end-user premises by one of System@tic, Paris region system maintenance ponents. the SALOME expert. versions every and ICT cluster. This includes for ex- 6 months. The analysis of simulation ample the projects IOLS, EHPOC, Assistance to create SALOME data, in particular using visuali- OpenHPC, ILMAB and OASIS. extension modules or solver inte- Bugtracking 500 fixes and zation of physical fields resulting gration. improvments / from computation workflows. Patch request for an immediate year. In this context, a key point of the access to correction, bug fixing Verification 4000 tests. platform is the usage of standard and intermediate certified re- & Validation data models to describe physical leases. concepts for numerical analysis, Users 300 users at EDF For more details, consult: and to ensure interoperability be- http://www.salome- and CEA, tween software components. The 4000 users over platform.org/service-and- MED data model for example is support/available-programs the world, used to describe geometric meshes 40000 downloads and physical fields. in year 2012 2 DEVELOPING DOMAIN SPECIFIC APPLICATIONS

The SALOME platform provides a software suite to integrate domain specific tools and realize dedicated solutions for simulation of physical systems. For example, you can download from http://www.code-aster.org the SALOME-MECA application that is the EDF simulation environment to investigate solid and structural mechanics questions about equipments ageing, reliability and safety of production plants. This application is a software configuration that integrate the SALOME platform with the Code_Aster solver and its computing environment.

A second example is PANTHEREV2, a numerical simulation application dedicated to industrial radiation protection. The screenshot below il- lustrates the computation of the average dose rate in a nuclear power plant (Figure 2).

Figure2: Evaluation of the average dose rate of radiation using the PANTHERE application (EDF/SEPTEN).

The following third example is a GUI interface for GENEPI+ code named PPGP and dedicated to the analysis of steam generators. It helps the user to describe the generators and its associated physical properties, and then manage the computing workflow from the meshing process to the data analysis and visualisation (Figure 3). In this context, the SALOME platform is distributed with a Quality Assurance industrial process, including a Verification and Validation activity. The development team provides first level services and support to investigate the development of dedicated applications by third party projects.

Figure3: PPGP application, at left: tree of a all elements of a steam generator, at middle: python access to specific changes, at right: dialog box to edit values of a part of the steam generator 3 MODELLING PHYSICAL SYSTEMS

The SALOME platform provides you with a whole set of modules to create complex geometric models. This includes high level meshing functionalities to prepare numerical models that fit the solver requirements and simulation objectives, in terms of accuracy and performance.

MODELLING MESHING THE GEOMETRY THE GEOMETRY The Geometry module (GEOM) The SALOME platform comes na- provides a rich set of commands to tively with a set of modules to cre- create, edit, import or modify a ate Finite-Elements representations complex CAD model. The geomet- of the geometry of your system (e.g. ric shapes may be designed interac- SMESH module ) and to refine tively using the Graphical User parts of the mesh representation Interface (GUI) or the Text User considering the solver specific re- Interface (TUI) through python quirements and the simulation ob- scripts. Each functionality of GEOM jectives (e.g. HOMARD module ). is available in TUI with facilities to The Meshing module work back and forth with the The Meshing module (SMESH) graphical interface. This allows to transforms the solid shapes defined build complex shapes or several or imported in the Geometry mod- configurations of a shape with dif- ule into finite-elements. The ferent values of its parameters. Meshing module is used to create The module is powered by a geom- and edit the mesh data using a va- etry kernel based on the Open riety of different open source or 3rd CASCADE Technology which pro- parties meshing algorithms. vides a Boundary representation of A concept of sub-meshes can be the model (BRep) and maintains used to take into account the spe- the topological structure required cific features of the geometrical by the subsequent meshing opera- model. A different set of conditions tions. can be applied to each sub-mesh, SALOME can exchanges geometry Figure 5: Representative Volume Elements generation for material for example to mesh with smaller studies (CEA/DEN). with other CAD systems using for- elements some area of the shape, mats as IGES, STEP, and BREP (the or to mesh an area with another Open CASCADE native format), kind of elements. (Figures 4 & 5). A complete toolbox enables the user to verify the mesh quality and to perform local modification or ad- justment. Transformation operations can be used to produce complex meshes or compounds. Mesh elements can be grouped to facilitate the visualization and help the definition of initial boundary conditions. These groups can be automatically deduced from the corresponding geometric groups, or generated using grouping crite- ria (named “filters” in SALOME). As for the CAD process, the meshing process can be entirely described into python scripts (The Text User Interface of SALOME) to handle complex studies or to ensure full reproducibility and parametrization of the simulation Figures 6&7: Crack modelling for ageing analysis of a a piping system. Left: external view of the workflow. (Figures 6 &7) complete crack. Right: internal view of the detailled meshing strategy 4 Figure 4: Geometry of a stator for magneto-thermal analysis (EDF/R&D).

Optimization and refinement Creating hexahedric models Figure 8: The Meshing module comes with a The SALOME platform provides you Mesh adaptation companion named the HOMARD with specific mesh algorithms that during the simulation of a module to perform specific mesh help you with the creation of com- tunnel excavation (EDF/R&D). adaptation to better fit the simula- plex geometric models with hexa- tion objectives, in term of accuracy hedric mesh representations. These and performance. meshes are specifically required for To improve the quality of the simu- some numeric solvers, for example lation results, mesh adaptation of- studies that involves fluid dynamic fers an effective compromise, solvers. combining a fine mesh with a low In this objectives, you computational cost. The HOMARD ® can use the HEXOTIC module allows refinement and plugin of the SMESH coarsening techniques to adapt the module. This plugin is a mesh, according to the numerical wrapper for the commer- error of the simulation. cial mesh program HOMARD ® is designed to operate MeshGems-Hexa (formerly in association with 2D/3D elements known as HEXOTIC) edited by the such as triangles, quadrangles, Distene Company. This function tetrahedrons and/or hexahedrons. can create automatically a volumic The whole mesh can be conformal hexahedric mesh of a complex or not. geometry without any cubic parti- Figure 9: Mesh representation of tion of the solids (Figure 9). a piping valve with automatic The selection of the elements to re- generation of hexahedron cells fine is made either by the local An alternative method is to use the (EDF/R&D). value of a field over the elements HEXABLOCK module that helps and a threshold, by a group or by a you to define a topological descrip- geometrical zone. Splitting their tion isomorphic to the real geome- edges in 2 refines these elements. try and from which a volumic The transition between different re- hexahedric mesh can be automati- finement zones is treated with spe- cally generated (Figure 10). cial elements. All HOMARD ® instructions can be executed either using the graphical user interface or through the python interface (Figure 8).

Figure 10: The topologic model of blocks (left), and the computed associated mesh (right) (CEA/DEN). 5 SUPERVISION OF COMPUTATION WORKFLOWS

There is an increasing need for multidisciplinary parametric simulations in various research and engineering domains. Fluid-structure interaction and thermal coupling are two examples. The software strategy in many contexts of simulation (at least at CEA and EDF) is to develop numerical solvers dedicated to their own domain, and then to execute multi-domains simulation by coupling the existing dedicated solvers (Figure 11).

Figure 11: Coupling of a neutronics model with a thermalhydraulics model for a nuclear safety study (CEA/DEN).

The SALOME platform provides you The YACS module contains Based on these core services dedi- with a set of services to create a the core set of services to com- cated to the workflows edition and simulation workflow that connects pose the computation workflow supervision, you will find high level different computation units and as a graph of connected services to help you in the design of then to execute the workflow on a SALOME components. a numeric experimental plan: distributed network of computers This graph can be edited using a The PARAMETRIC module and HPC resources: graphical user interface (GUI) or can be used to generate a para- The YACSGEN program helps the python Text User Interface metric simulation, by exploration you with the generation of a (TUI) to handle complex work- of a specified domain of the sim- SALOME component that inte- flow into scripts. ulation parameters space. grates your specific solver or any The JOBMANAGER module processing function. can be used to define a compu- The SALOME components are tation job (including either a sim- the computation units of a work- ple SALOME component or a flow designed with YACS. complete YACS workflow) and to drive the submission of the job to a distributed set of computers or HPC resources. The module can handle the batch systems PBS, LSF, SGE, LOADLEVELER or SLURM through a normalized generic interface. It comes with a graphical user interface, but can be used at a pro- gramming level using a python interface.

6 ANALYSIS OF SIMULATION DATA

PROCESSING FIELDS VISUALIZATION OF AND MESHES SIMULATION RESULTS In SALOME, the MED data model The solvers interfacing with defines a normalized representa- SALOME generate results that can PROCESSING tion to describe meshes of the be analysed within the ParaViS geometry and fields of the physical module. This module has been de- SIMULATION values of the simulation. This data veloped by integrating ParaView DATASET Figure 12: Projection of a field model is a key feature of the plat- (third party software edited by Beyond the processing and visuali- from a regular grid towards an form. It comes with a software im- Kitware on basis of the VTK toolkit) zation of meshes and fields data, unstructured mesh. plementation (the MED-file library) into SALOME, and exposes all the the SALOME platform contains for file persistence and intercom- functionalities of this award-win- additional modules dedicated to munication of simulation data be- ning post-processor tool. advanced data analysis : tween components. SALOME A wide range of representations are Data assimilation and opti- meshing and visualization compo- available to the physicists to explore misation environment , for nents propose interfaces with MED the datasets: surface, volume, example to recalibrate the format to import or export data. gauss points... The data can then parameters of a model by com- Domain specific solvers that inter- be analysed by using one of many parison of the simulation data to act with SALOME are advised to filters to extract significant data: clip, experimental measures (module use MED format for input and out- threshold, iso-surface, stream lines, ADAO) put files. This configuration leads to elevation surface… Propagation of uncertain- a high level of associativity between Quantitative information can be ex- ties in the simulation workflow, the software components to build tracted using the data analysis for example to evaluate the simulation workflows. tools: taking a selection of the data, uncertainty on resulting data of Based on the concepts from the histograms, 2D plots of time-plots the simulation considering a MED data model, the MED- are easily obtained. given uncertainty on the input Coupling module of SALOME All these features can be animated parameters (module OPEN- provides you with a full set of serv- TURNS). ices for high performance data pro- within the module to analyze time- Design numeric experimen- cessing on meshes and fields. varying data, sweep a cutting plane tal plans , for example a para- Extensively rewritten for SALOME 7, through the dataset, or animate a metric simulation by exploration the module contains the data struc- modal analysis. of a specified domain of the ture to describe meshes and fields This module is fully scriptable in simulation parameters space as C++ or python objects (MED- python to create visualizations in (module PARAMETRIC). Coupling package). It provides a set batch when necessary or to repeat of functions to manage the persist- analysis on ensemble runs, and can ence toward the med file format use visualization clusters (MEDLoader package), and func- to interactively tions to process the field data in analyse large standard algebraic operations datasets. (post-processing) or through inter- (Figure 13). polation and localization algorithms (INTERP_KERNEL and REMAPPER packages), for example to perform field projections from a mesh to another (Figure12).

Figure 13: Analysis of a Representative Volume Element in material studies (CEA/DEN).

7 Paris COM & Paris D - apital de 924 433 331 euros - 552 081 317 R.C.S. R.C.S. 317 081 552 - euros 331 433 924 de apital

CONTACT US: - Le Groupe EDF est certifié ISO 14001 - SA au c au SA - 14001 ISO certifié est EDF Groupe Le - Francis Kloss [email protected] Guillaume Boulant [email protected] Mikhail Kazakov [email protected] @

Commissariat à l’Energie Atomique Open Cascade et aux Energies Alternatives, EDF R&D Filiale d’EURIWARE, groupe AREVA, CEA-Saclay, DEN, DM2S, 1, avenue du Général de Gaulle, 1 place des frères Montgolfier 91191 Gif-sur-Yvette Cedex, France 92141 Clamart Cedex, France 78044 Guyancourt, France

www.cea.fr www.edf.com www.opencascade.com DÉVELOPPEMENT ET RECHERCHE EDF 2013 Novembre -