Towards overcoming the LES crisis. Issue 3 (16th March 2019)

• Record Type: Journal Article
• Authors: Löhner, Rainald
• Is Part Of: International journal of computational fluid dynamics. Volume 33:Issue 3(2019)
• Page Start: 87
• Page End: 97
• Publication Date: 2019-03-16
• Subjects: Large eddy simulation -- CFD -- HPC -- finite difference methods -- Lattice–Boltzmann methods -- industrial LES; Fluid dynamics -- Data processing -- Periodicals; 532.05
• DOI: 10.1080/10618562.2019.1612052
• Languages: English
• ISSNs: 1061-8562

Abstract

ABSTRACT: Industrial large-eddy simulation (LES), i.e. overnight runs with O ( 10 9 ) degrees of freedom (DOFs) and O ( 10 7 ) timesteps, has been one of the top priorities of CFD research over the last two decades. Current network and solver timings indicate that the required target of 5 ms/timestep is within reach: some cumulant Lattice Boltzmann Method (LBM) codes are reporting speeds in this range using multi-GPU systems, while some Finite Difference Method (FDM) codes running on traditional CPUs are achieving speeds that are only a factor of 2–4 slower. This implies that the long wait may be coming to a close. During the last three decades, a number of promising approaches have been tried. Given that the majority of these were promoted as the 'solution to LES' or the 'solution to turbulence', the paper lists them under the label of 'the false prophecies'. Furthermore, some of the assumptions that are always part of the scientific discovery process turned out to be incorrect. These are listed under the term 'the false assumptions'. From an informal survey conducted in January of 2018, it appears that simple Cartesian Finite Difference codes or, equivalently, Lattice Boltzmann codes may be the first to achieve industrial LES.