Large-scale Simulations in Stability, Transition, Turbulence and Control
||Large-scale Simulations in Stability, Transition, Turbulence and Control|
||SNIC Large Compute|
||Philipp Schlatter <firstname.lastname@example.org>|
||Kungliga Tekniska högskolan|
||2022-01-01 – 2023-01-01|
||20306 10508 10501|
We present a large-level request for computer time on high-performance computing (HPC) resources within the Swedish National Infrastructure for Computing (SNIC). The proposed projects by the research groups of the KTH Engineering Mechanics department are summarized. The group of applicants consists of a total of 7 senior researchers, 4 application experts, 6 Postdocs and 16 PhD students, i.e. a total of 33 researchers. We actively promote collaboration within our large user group to facilitate HPC support, sharing of simulation methods, codes, data, post-processing, data management methods and user experience. We have thus found it benecial to apply for a large-level allocation instead of multiple medium-level requests.
In this proposal we describe our scientic projects which rely on HPC resources, grouped into four main areas: 1) aeronautics; 2) wind turbines and geophysical ows; 3) dynamical systems, uncertainty quantication and machine learning; 4) fundamentals of transition and turbulence. Due to the large number of collaborators, we do not list all individual projects, but rather give an overview of the general research directions.
Our research makes use of the numerical codes described in Section 2 below, and the specic data management plan described in complementary storage application. Note that we get specic application support through the Swedish e-Science Research Centre (SeRC) and EuroHPC competence centre in the form of four application experts and we actively develop our codes. In particular, the possibility to use different machines depending on job size and job characteristic is benecial for the efficient usage of the available computer time; also having a good mix between computer centres has been helpful to us. We have thus found that, depending on problem size and code, Tetralith (NSC), Kebnekaise (HPC2N), Dardel (PDC) and LUMI (CSC) are excellent choices.