Experimental work has been performed to investigate turbulent Couette- type flow with
repeated longditudinal pressure gradients that are realized in a channel consisting of a
wavy fixed wall on one side and a moving wall on the other.
This type of flow simulates the flow in high speed journal bearings operating in turbulent
conditions.
2D flow with constant temperature.

The channel has its lower wall moving and its upper wall wavy and fixed.
The test section is L=1.4 m long (one wavelength) with a mean height (distance
between the two walls) of: 2
=15 mm.
Equation of the fixed wavy wall: y=2
-e
cos(2
x/L) with e=4.5 mm

The figure shown above describes the test channel and co-ordinate system used. The
measurements have been carried out in the middle part of the channel (total length of 3.5
times the wavelength) in order to minimize upstream and downstream effects. The pressure
has been found to be the same at the inlet and outlet of the considered one
wavelength-long portion. Moreover the longitudinal pressure gradient is also found to be
periodic.
The mean velocity distributions and turbulent intensities are measured at various
streamwise positions of x/l for Reynolds number of Re=5000 and 8000. Non-dimensional
pressure gradient parameters
![]()
can be seen in the table below.

Channel flow at x=0. Velocity measurements stored in file couette.prn
Profiles from fixed wall to moving wall of:
/U
,
/U
,
/U
,
,
/U
² All velocity measurements were made using hot-wire anemometer.
Reynolds shear stresses were measured by hot-wire anemometer.
Measurement errors are not published in any English language paper.
/U
,
/U
,
/U
,
,
/U
²
/dx Ideally the calculation of the channel flow should be performed using periodicity
conditions for both velocities and pressure gradient (d
(x+L)/dx=d
(x)/dx).
Nevertheless it is also possible to start at x=0 prescribing the experimental
values provided as inlet conditions and assume zero gradients for the flow variables at
outlet. In this case the flow periodicity has to be checked.
The following results should be plotted and compared with the data:
and d
/dx distributions along
the wavy wall. 