Search

The paper presents new approach to Static and Dynamic Load Balancing (DBL) for loosely coupled multiscale modelling methods executed in heterogeneous hardware infrastructure. The most popular configurations of computing nodes composed of modern multicore CPUs, GPUs and co-processors, are used. The proposed load balancing approach takes into account previously determined computational character of methods applied in particular scales, which depends on a size of input data, operational intensity and limitations of hardware architecture. Such limitations are defined by the Roofline model and then used in the algorithm as boundary conditions, allowing to determine maximum performance of an algorithm for particular device. Multiscale calculations based on upscaling approach are analysed by using Cellular Automata Finite Element (CAFE) method. The qualitative results, obtained after application of proposed load balancing procedure, are discussed in the paper in details.

Modelling and simulation of multiscale systems constitutes a grand challenge in computational science, and is widely applied in fields ranging from the physical sciences and engineering to the life sciences and the socio-economic domain. To adequately simulate numerous intertwined processes characterized by different spatial and temporal scales (often spanning many orders of magnitude), sophisticated models and advanced computational techniques are required. Additionally, these multiscale models frequently need large scale computing capabilities as well as dedicated software and services that enable the exploitation of existing and evolving computational ecosystems. The aim of the annual Workshop on Multiscale Modelling and Simulation is to facilitate the progress in this multidisciplinary research field http://www.computationalscience.nl/MMS/. In this paper, we reflect on the 12 years of workshop history and glimpse at the latest developments presented in 2015 in Iceland, the Land of Fire and Ice. In Section 6, we invite new workshop co-organizers.

Multiscale Modelling and Simulation (MMS) is a cornerstone in today’s research in computational science. Simulations containing multiple models, with each model operating at a different temporal or spatial scale, are a challenging setting that frequently require innovative approaches in areas such as scale bridging, code deployment, error quantification, and scientific analysis. The aim of the MMS workshop is to encourage and consolidate the progress in this multi-disciplinary research field, both in the areas of the scientific applications and the underlying infrastructures that enable these applications. Here we briefly introduce the scope of the workshop and highlight some of the key aspects of this year’s submissions.

Ab initio simulations aimed at modelling and interpreting STM induced molecular isomerization of naphthalocyanine molecule are presented. Free energy profile, reaction path and activation energies were obtained using the metadynamics method in the frame of Car-Parrinello Molecular Dynamics (CPMD). We propose a multiscale model of the molecular switch process, based on the Ehrenfest Molecular Dynamics and on the knowledge of the electronic excited states as computed from ab initio calculations. The time evolution of the electronic density during the switch process induced by an external bias potential is calculated exploiting the multiscale model. Simulations were performed on IBM Blue Gene/P supercomputer at Moscow State University.

Severe events around the globe have highlighted the threat to life, infrastructure and the environment posed by flooding. Flood forecasting systems are a vital component of broader flood risk management activities. These systems are becoming increasingly more sophisticated as their importance in reducing life loss and economic damages is realized. Part of this increase in complexity is focused on the ability to predict and warn of failures in dykes, levees and embankments. A new European ICT project, UrbanFlood for Environmental Services and Climate Change Adaptation, has recently been commissioned and is introduced in this presentation. The primary objective of the Urban Flood project is to develop early warning systems that will monitor flood protection systems in real-time, identify vulnerable locations, model the failure and predict dike collapse and subsequent inundation. In combination with the damage assessment, Urban Flood will serve as an advanced decision support system, mitigating the impact of seasonal and catastrophic floods. Modeling is one of the key tasks in the project. The models will be required to simulate the behavior of the material properties of the layered dikes (sand, clay, peat, grass or concrete cover, metal frame, dam gates, etc.), during extreme hydraulic loading events. In earthen dikes, extra challenge is posed by the non-linear elastic plastic properties of the deformable clay. A realistic simulation of the dike will model the free-surface water dynamics; convective and diffusive transfer of water inside the porous materials; dynamic response of clay to the water pressure; structural mechanics, deformation and actual dike breakdown and flood. The models shall cover a wide range of scales from a sand grain to a flooded city. The time scales will range from seconds (for water penetrating the soil) to hours (for dike collapse dynamics and ocean tides). Eventually, the models will predict the influence of seasonal and global changes on the stability of flood defense systems. Full 3D transient simulation of dike failure with subsequent inundation will require significant computing resources. The project started three months ago, and we will present the plan for developing the modeling cascade for the system. This work is supported by the UrbanFlood European Union project N 248767, theme ICT-2009.6.4

Statistically Similar Representative Volume Element (SSRVE) is used to simplify computational domain for microstructure representation of material in multiscale modelling. The procedure of SSRVE creation is based on optimization loop which allows to find the highest similarity between SSRVE and an original material microstructure. The objective function in this optimization is built upon computationally intensive numerical methods, including simulations of virtual material deformation, which is very time consuming. To avoid such long lasting calcu- lations we propose to use the data farming approach to identification of SSRVE for Advanced High Strength Steels (AHSS) characterized by multiphase microstructure. The optimization method is based on a nature inspired approach which facilitates distribution and parallelization. The concept of SSRVE creation as well as the software architecture of the proposed solution is described in the paper in details. It is followed by examples of the results obtained for the identification of SSRVE parameters for DP steels which are widely exploited in modern automotive industry. Possible directions for further development as well as possible industrial applications are described in the conclusions.

Nearly all systems of practical interest are composed of parts assembled across multiple scales. For example, an agrodynamic system is composed of flora and fauna on one scale; soil types, slope, and water runoff on another scale; and management practice and yield on another scale. Or consider an advanced coal-fired power plant: combustion and pollutant formation occurs on one scale, the plant components on another scale, and the overall performance of the power system is measured on another. In spite of this, there are few practical tools for the optimization of multiscale systems. This paper examines multiscale optimization of systems composed of discrete elements using the plus-one-recall-store (PORS) problem as a test case or study problem for multiscale systems. From this study, it is found that by recognizing the constraints and patterns present in discrete multiscale systems, the solution time can be significantly reduced and much more complex problems can be optimized.

This paper presents a systematic survey of open source multiphysics frameworks in the en- gineering domains. These domains share many commonalities despite the diverse application areas. A thorough search for the available frameworks with both academic and industrial ori- gins has revealed numerous candidates. Considering key characteristics such as project size, maturity and visibility, we selected Elmer, OpenFOAM and Salome for a detailed analysis. All the public documentation for these tools has been manually collected and inspected. Based on the analysis, we built a feature model for multiphysics in engineering, which captures the commonalities and variability in the domain. We in turn validated the resulting model via two other tools; Kratos by manual inspection, and OOFEM by means of expert validation by domain experts. Keywords: Multiphysics; Multiscale; Modelling and Simulation; Domain Analysis; Feature Model