Application of a JPEG 2000-Based Data Compression Algorithm to DNS of Compressible Turbulent Boundary Layers Up to $$Re_\theta =6600$$ R e θ = 6600

Christoph Wenzel (), Patrick Vogler (), Johannes M. F. Peter (), Markus J. Kloker () and Ulrich Rist ()
Additional contact information
Christoph Wenzel: Universität Stuttgart, Institut für Aero- und Gasdynamik (IAG)
Patrick Vogler: High Performance Computing Center (HLRS), Stuttgart
Johannes M. F. Peter: Universität Stuttgart, Institut für Aero- und Gasdynamik (IAG)
Markus J. Kloker: Universität Stuttgart, Institut für Aero- und Gasdynamik (IAG)
Ulrich Rist: Universität Stuttgart, Institut für Aero- und Gasdynamik (IAG)

A chapter in High Performance Computing in Science and Engineering '20, 2021, pp 295-313 from Springer

Abstract: Abstract This paper presents a performance analysis of the compression library BigWhoop applied to compressible-turbulent-boundary-layer data obtained by direct numerical simulation (DNS). The DNS data are computed for a Mach number of $$M=2.0$$ M = 2.0 and cover a Reynolds-number range of $$Re_\theta =300$$ R e θ = 300 to 6600. Including six flow-field variables, each time step has a data size of about 1 TB. Mainly evaluated for the streamwise velocity component, various compression rates are tested for the flow field of one single time step as data post processing. It is shown that DNS data compressed up to 1 : 200 yield a maximum absolute error of $$0.2\%$$ 0.2 % for the entire domain. This is still accurate enough to reliably perform flow-field investigations in the outer layer of the turbulent boundary layer, notably considering that the maximum error only occurs with extremely low probability. The no-slip wall boundary condition forces the velocity down to zero, and thus the relative error, i.e. the absolute error divided by the local velocity, is significantly larger in the near-wall region. Hence, for flow-field investigations depending on important wall values like the skin friction for instance, the maximum compression ratio should be limited to about 1 : 100, keeping the maximum relative error below $$1\%$$ 1 % in the near-wall region.

Date: 2021
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:sprchp:978-3-030-80602-6_19

Ordering information: This item can be ordered from
http://www.springer.com/9783030806026

DOI: 10.1007/978-3-030-80602-6_19

Access Statistics for this chapter

More chapters in Springer Books from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().