Data Bank contribution: P. Bradshaw, B.E. Launder and J.L.
Lumley, "Collaborative Testing of Turbulence Models", J.
Fluids Engg vol. 118, pp. 243-247, June 1996
This is CTTM data library disk 0 - introduction and
documents. All files in this data bank contribution are in
ASCII.
CASES15 TXT 31132 01-14-97 12:19p
ARCH1 TXT 54072 01-14-97 1:24p
CTTMFIN TEX 51447 10-04-96 1:29p
FIG1 EPS 13118 01-14-97 1:54p
CTTMFIN PS 185700 01-14-97 1:50p
8 file(s) 367644 bytes
1088512 bytes free
File ARCH1.TXT is the main description of the library. It
contains directory listings for the "data disks" (now
subdirectories) D1-D6 and R1-R7, and a table of
contents of the data library of the 1980/81 Stanford
Conference on Computation of Complex Turbulent Flows, which
occupies disks D1-D3 with some material on later disks. It
also contains detailed running instructions, as issued to
the Collaborators, for the "complex" flows on Disk D6:
instructions for the simpler cases on Disks D1 to D5 (only
the cases actually used in the Collaboration) are in file
CASES15.TXT.
CTTMFIN is the final report to the funding agencies,
recorded in TeX (not LaTeX) format and also as a PostScript
file. The one figure is - necessarily - in a PostScript
file, FIG1.EPS.
The CTTM project began with "entry" test cases, namely the
prediction of skin-friction coefficient in a boundary layer
in zero pressure gradient at a momentum-thickness Reynolds
number of 10000. The consensus value in low-speed flow is
0.00264, with an uncertainty of no more than +/- 2 percent
(the entry cases also included a Mach 5 adiabatic-wall
boundary layer and two boundary layers with heat transfer).
The "entry" test cases proved invaluable in spotting errors
in models or codes. We most strongly recommend that any
turbulence model should first be tested in a low-speed
constant-pressure boundary layer, and the predicted skin-
friction coefficient and logarithmic-law constants checked
against standard data. Grid spacing requirements in the
near-wall region can also be checked in this simple flow.
Codes which converge only at finite Mach number can be run
at, say, M=0.3 and M=0.5, and the results extrapolated back
to M=0 on the assumption that compressibility effects vary
as the square of M.