| 1. |
- Joffrin, E., et al.
(författare)
-
Overview of the JET preparation for deuterium-tritium operation with the ITER like-wall
- 2019
-
Ingår i: Nuclear Fusion. - : IOP Publishing. - 1741-4326 .- 0029-5515. ; 59:11
-
Forskningsöversikt (refereegranskat)abstract
- For the past several years, the JET scientific programme (Pamela et al 2007 Fusion Eng. Des. 82 590) has been engaged in a multi-campaign effort, including experiments in D, H and T, leading up to 2020 and the first experiments with 50%/50% D-T mixtures since 1997 and the first ever D-T plasmas with the ITER mix of plasma-facing component materials. For this purpose, a concerted physics and technology programme was launched with a view to prepare the D-T campaign (DTE2). This paper addresses the key elements developed by the JET programme directly contributing to the D-T preparation. This intense preparation includes the review of the physics basis for the D-T operational scenarios, including the fusion power predictions through first principle and integrated modelling, and the impact of isotopes in the operation and physics of D-T plasmas (thermal and particle transport, high confinement mode (H-mode) access, Be and W erosion, fuel recovery, etc). This effort also requires improving several aspects of plasma operation for DTE2, such as real time control schemes, heat load control, disruption avoidance and a mitigation system (including the installation of a new shattered pellet injector), novel ion cyclotron resonance heating schemes (such as the three-ions scheme), new diagnostics (neutron camera and spectrometer, active Alfven eigenmode antennas, neutral gauges, radiation hard imaging systems...) and the calibration of the JET neutron diagnostics at 14 MeV for accurate fusion power measurement. The active preparation of JET for the 2020 D-T campaign provides an incomparable source of information and a basis for the future D-T operation of ITER, and it is also foreseen that a large number of key physics issues will be addressed in support of burning plasmas.
|
|
| 2. |
- Wiesenberger, M., et al.
(författare)
-
Reproducibility, accuracy and performance of the FELTOR code and library on parallel computer architectures
- 2019
-
Ingår i: Computer Physics Communications. - : Elsevier. - 0010-4655 .- 1879-2944. ; 238, s. 145-156
-
Tidskriftsartikel (refereegranskat)abstract
- FELTOR is a modular and free scientific software package. It allows developing platform independent code that runs on a variety of parallel computer architectures ranging from laptop CPUs to multi-GPU distributed memory systems. FELTOR consists of both a numerical library and a collection of application codes built on top of the library. Its main targets are two- and three-dimensional drift- and gyro-fluid simulations with discontinuous Galerkin methods as the main numerical discretization technique. We observe that numerical simulations of a recently developed gyro-fluid model produce non-deterministic results in parallel computations. First, we show how we restore accuracy and bitwise reproducibility algorithmically and programmatically. In particular, we adopt an implementation of the exactly rounded dot product based on long accumulators, which avoids accuracy losses especially in parallel applications. However, reproducibility and accuracy alone fail to indicate correct simulation behavior. In fact, in the physical model slightly different initial conditions lead to vastly different end states. This behavior translates to its numerical representation. Pointwise convergence, even in principle, becomes impossible for long simulation times. We briefly discuss alternative methods to ensure the correctness of results like the convergence of reduced physical quantities of interest, ensemble simulations, invariants or reduced simulation times. In a second part, we explore important performance tuning considerations. We identify latency and memory bandwidth as the main performance indicators of our routines. Based on these, we propose a parallel performance model that predicts the execution time of algorithms implemented in FELTOR and test our model on a selection of parallel hardware architectures. We are able to predict the execution time with a relative error of less than 25% for problem sizes between 10 −1 and 10 3 MB. Finally, we find that the product of latency and bandwidth gives a minimum array size per compute node to achieve a scaling efficiency above 50% (both strong and weak).
|
|
