STATISTICAL ANALYSIS OF WALL, TURBULENCE PHENOMENA
25 1
2. EXPERIMENTAI. TECl4NIQtJE
The flow configuration is a simple one. Measurements have been made in
air flow at the straight outlet section of a channel used previously for other
experiments and described elsewhere (Zaric, 1972a). A fairly low Reynolds
number of Re = 39.000 was chosen in order to thicken the viscous sublayer.
Maximum velocity in the 40 by 300 mm cross section was 10.2 m/s. with the
friction velocity amounting to u* = 0.48 m/s.
Hot wire anemometry is employed as the most suitable existing technique
for statistical analysis, in spite of a number of disadvantages when used in
the wall vicinity. The pronounced nonlinearity of the signal at very low
velocities appearing in the viscous sublayer requires very careful calibration
and digital linearization. The wall effect requires special corrections which
are still mainly empirical. At high turbulence intensities in the viscous sublayer the fact that the wire is equally sensitive to the two velocity vector
components normal to the wire has to be taken into account.
A single, 5-pm diameter, tungsten wire is employed in order to approach
the wall as close as possible. The wire could be switched either to a DISA
anemometer, or to a Mueller bridge so that the flow temperature could also
be detected. More details on the technique are given elsewhere (Zaric,
1972a). Signals from the anemometer are registered on an analogue tape
running at 1.524 m/s, using an AMPEX FR-1300 tape recorder operated in
the FM mode. The tape is replayed with a 1 : 32 speed reduction and the
signal is fed into the digital computer with a frequency of 250 Hz, via an
analogue-todigital conversion system. The real time sampling frequency
being 8O00 Hz, 15-s long signals are registered on a digital tape in the form
of a succession of 120,000 instantaneous values. The statistical analysis is
performed on a CDC-360 digital computer with a 64K memory.
3. CONVENTIONAL STATISTICAL ANALYSIS
Feeling that amplitude statistical analysis was somehow neglected in wall
turbulence research, we have emphasized probability density distribution
analysis. Velocity probability densities are determined on the basis of
120,000 instantaneous values for 20 different nondimensional distances from
the wall, ranging from Y
+
= 1.6 to Y
+
= 21 1. Complete probability density
distributions have been presented elsewhere (ZariC, 1972~). Probability density distributions in the buffer layer only are presented in Fig. 1. In Fig. 2
distributions of the turbulence intensities KIWI, skewness factors S, and
flatness factors F are presented as functions of the distance from the wall.
25 1
2. EXPERIMENTAI. TECl4NIQtJE
The flow configuration is a simple one. Measurements have been made in
air flow at the straight outlet section of a channel used previously for other
experiments and described elsewhere (Zaric, 1972a). A fairly low Reynolds
number of Re = 39.000 was chosen in order to thicken the viscous sublayer.
Maximum velocity in the 40 by 300 mm cross section was 10.2 m/s. with the
friction velocity amounting to u* = 0.48 m/s.
Hot wire anemometry is employed as the most suitable existing technique
for statistical analysis, in spite of a number of disadvantages when used in
the wall vicinity. The pronounced nonlinearity of the signal at very low
velocities appearing in the viscous sublayer requires very careful calibration
and digital linearization. The wall effect requires special corrections which
are still mainly empirical. At high turbulence intensities in the viscous sublayer the fact that the wire is equally sensitive to the two velocity vector
components normal to the wire has to be taken into account.
A single, 5-pm diameter, tungsten wire is employed in order to approach
the wall as close as possible. The wire could be switched either to a DISA
anemometer, or to a Mueller bridge so that the flow temperature could also
be detected. More details on the technique are given elsewhere (Zaric,
1972a). Signals from the anemometer are registered on an analogue tape
running at 1.524 m/s, using an AMPEX FR-1300 tape recorder operated in
the FM mode. The tape is replayed with a 1 : 32 speed reduction and the
signal is fed into the digital computer with a frequency of 250 Hz, via an
analogue-todigital conversion system. The real time sampling frequency
being 8O00 Hz, 15-s long signals are registered on a digital tape in the form
of a succession of 120,000 instantaneous values. The statistical analysis is
performed on a CDC-360 digital computer with a 64K memory.
3. CONVENTIONAL STATISTICAL ANALYSIS
Feeling that amplitude statistical analysis was somehow neglected in wall
turbulence research, we have emphasized probability density distribution
analysis. Velocity probability densities are determined on the basis of
120,000 instantaneous values for 20 different nondimensional distances from
the wall, ranging from Y
+
= 1.6 to Y
+
= 21 1. Complete probability density
distributions have been presented elsewhere (ZariC, 1972~). Probability density distributions in the buffer layer only are presented in Fig. 1. In Fig. 2
distributions of the turbulence intensities KIWI, skewness factors S, and
flatness factors F are presented as functions of the distance from the wall.