Search

The next generation of tokamaks, e.g. ITER, will have extremely large data collection rates significantly larger than those experienced today in present tokamaks, with consequential new challenges in data management, data analysis and integrated modelling. One of these challenges is to ensure that appropriate data is efficiently made available when it is required and where it is consumed. Data volumes with limited network capabilities mean not all data can be distributed in time when a data-object is requested. One possible solution is to preemptively identify and distribute efficiently the data across the storage services before a user or an application requests it. Preemptive data distribution rely on analysis of historical access patterns to identify a set of rules whereby following a data-object request the most probable set of next requests can be inferred. Implementation of these rules requires the inferred sets of data to be moved close to data-object consumer. The work presented will describe the Apache Spark Machine Learning tools, the results of the analysis, and an implementation of the preemptive distribution experimental platform at CCFE, together with plans for its future integration and testing on the upcoming SAGE platform.

Journal publications, as the final product of research activity, are the result of an extensive complex modeling and data analysis effort. It is of paramount importance, therefore, to capture the origins and derivation of the published data in order to achieve high levels of scientific reproducibility, transparency, internal and external data reuse and dissemination. The consequence of the modern research paradigm is that high performance computing and data management systems, together with metadata cataloguing, have become crucial elements within the nuclear fusion scientific data lifecycle. This paper describes an approach to the task of automatically gathering and cataloguing provenance metadata, currently under development and testing at Culham Center for Fusion Energy. The approach is being applied to a machine-agnostic code that calculates the axisymmetric equilibrium force balance in tokamaks, EFIT++, as a proof of principle test. The proposed approach avoids any code instrumentation or modification. It is based on the observation and monitoring of input preparation, workflow and code execution, system calls, log file data collection and interaction with the version control system. Pre-processing, post-processing, and data export and storage are monitored during the code runtime. Input data signals are captured using a data distribution platform called IDAM. The final objective of the catalogue is to create a complete description of the modeling activity, including user comments, and the relationship between data output, the main experimental database and the execution environment. For an intershot or post-pulse analysis (∼1000 time slices, 65 × 65 grid, mpi execution n = 8 cores) of a typical MAST pulse, the overhead in the code runtime caused by the Provenance Metadata Gathering System is less than 10%, the metadata/data size ratio is about ∼20%, which we consider to be reasonable according to the present literature. A visualization interface based on Gephi for catalogue interrogation, will be presented.

Intense thermal loads in fusion devices occur during plasma disruptions, Edge Localized Modes (ELM) and Vertical Displacement Events (VDE). They will result in macroscopic erosion of the plasma facing materials and consequent accumulation of activated dust into the ITER Vacuum Vessel (VV). A recognized safety issue for future fusion reactors fueled with deuterium and tritium is the generation of sizeable quantities of dust. In case of LOVA, air inlet occurs due to the pressure difference between the atmospheric condition and the internal condition. It causes mobilization of the dust that can exit the VV threatening public safety because it may contain tritium, may be radioactive from activation products, and may be chemically reactive and/or toxic (Sharpe et al. [1] ; Sharpe and Humrickhouse [2] ). Several experiments have been conducted with STARDUST facility in order to reproduce a low pressurization rate (300 Pa/s) LOVA event in ITER due to a small air leakage for two different positions of the leak, at the equatorial port level and at the divertor port level, in order to evaluate the velocity magnitude in case of a LOVA that is strictly connected with dust mobilization phenomena. A two-dimensional (2D) modelling of STARDUST, made with the CFD commercial code FLUENT, has been carried out. The results of these simulations were compared against the experimental data for CFD code validation. For validation purposes, the CFD simulation data were extracted at the same locations as the experimental data were collected. In this paper, the authors present and discuss the computer-simulation data and compare them with data collected during the laboratory studies at the University of Rome “Tor Vergata” Quantum Electronics and Plasmas lab.