Index of /turbdata/agard/chapter5/PCH20
Title: Fully Developed Rotating Channel Flow (Johnston, Hallen & Lezius)
Description: This is the README file with general information regarding
the data available for PCH20. Included is information on file
organization, definitions, and uncertainties.
---End of Header---
Prof. James P. Johnston, Mechanical Engineering Dept.
Stanford University, Stanford, CA 94305-3030
(1) R. M. Halleen (1967), "The Influence of Rotation on Flow in a Long
Rectangular Channel - an Experimental Study," PhD dissertation, Department
of Mechanical Engineering, Stanford University, Stanford, CA 94305, June
(Source of original data -- request from Stanford University Library, or
order from University Microfilms.)
(2) J. P. Johnston, R. M. Halleen and D. K. Lezius (1972), "Effects of
Spanwise Rotation on Structure of Two-Dimensional, Fully Developed
Turbulent Channel Flow," Journal of Fluid Mechanics, Vol. 56, p. 533-577.
(3) J. P. Johnston (1973), "The Suppression of Shear Layer Turbulence in
Rotating Systems," Trans. ASME, Journal of Fluids Engineering, Vol. 95,
Series I, No. 2, pp. 229-236.
Re = 2h*Um/nu, Reynolds number
Ro = 2h*Omega/Um, Rotation number
h = Channel 1/2 width (Note: D in references)
Um = Integrated mean velocity for a profile, see ref. (2).
Utaw = Wall shear velocity
Utawo = Wall shear velocity at Ro = 0 with Re number held constant
nu = kinematic viscosity
Omega = rotation rate, axis perpendicular to x-y plane
x = distance downstream of channel inlet
y = distance away from trailing (stable) wall of channel (Note: y has
opposite sign from values used in references, see Fig. 1, ref. (2))
The data files from the original source, ref. (1), were recompiled and
reanalysed. A limited set of mean velocity profiles at two Reynolds
numbers are presented together with their dimensionless wall shear
velocities on leading (unstable) and trailing (stable) side walls.
The geometrical configuration is shown in Fig. 3 of ref. (2). The channel
was 59 inches long, 11 inches between end walls (along axis of rotation)
and 1.54 inches between side walls (2h = 1.54 inches). The fluid was water
at room temperature. Mean velocity profiles were measured in the central
plane, midway between end walls at two positions, x = 52.4 inches and 44.7
inches downstream of the channel inlet. For the high Reynolds number case,
comparison of upstream and downstream profiles indicates that the flow may
not be quite fully developed. Profiles at the downstream station, x/h = 68
are presented here. Some of these profiles were plotted in Fig. 14 of ref.
Ref. (2) gives a brief account of the experimental methods which are
discussed in detail in ref. (1). Experimental error in the results was
accessed and the values given below are generous (low) estimates,
considering the techniques that were employed at the time the original data
The dimensionless mean velocities have an uncertainty of + or - 1.5 per
cent of the profile mean. The absolute error in tabulated dimensionless
velocity, (U/Um) = + or - 0.02.
The wall shear velocities have a higher uncertainty, about + or - 2.5 per
cent of the value at zero rate of rotation. The absolute error in the
tabulated dimensionless ratios, (Utaw/Utawo) = + or - 0.06.
Three files were prepared for final distribution in 1996.
pro35k.txt - Mean velocity profiles at Re = 35,000 (five profiles with Ro =
0 to 0.08)
pro11k.txt - Mean velocity profiles at Re = 11,000 (five profiles with Ro =
0 to 0.21)
utaws.txt - Wall shear velocity ratios for both sets of velocity profiles.
Each profile has an ID#, PXXYYZZ. XX = x/h. YY = nominal Re. ZZ = nominal
I believe that use of these data for comparison to modern CFD results
is only warranted for qualitative purposes, considering their
experimental uncertainty. In addition, the higher Re flow (Re =
35,000) may not be quite fully developed. The methods available in
the late 1960's were not as inherently accurate as modern experimental
methods, and we possessed no instrumentation, at that time, for
measurement of the turbulent Reynolds stresses, quantities which
should be checked against CFD. If this flow continues to be an
important test case, it is recommended that new and better experiments
James P. Johnston