n,,~
by
accumulating
botlndary
Izyer material from the pressure fncc of the channel
in
,jl,
boundary layer on the suction face. For basic considcrations on
the
gencrntion of
secondary flow, set. also the report of
J.
D.
Johnston
5).
In the channel of a rotor the cffccl due to Coriolis force i~sually prcdorninates. Hencc thc boundary
I;lyel. fiuid from the pressure face is accu~nulatcd along the boundary layer of the suction face iow;lrds
the
outlet of this channel. This phenomenon was observed by colour injection. Iiigh speed calncrn.
rotoscopc alld
transparent
shroud by
Fischer
and
Thornn
in
1930
15.561
in
a
pump impeller.
11
was rcconfirmcd
by
means of hot wire anemometry in a str;tight rotating channcl wit11
rectangular
cross section by
hloort
at
MIT
in
1962
[5.57].
The measurcmcnts of hfooi~
how
also thc rctclltion
of
the logarithmic velocity distribution at the wall in the decisive loss-generating zones at pressure
suction face.
~lthotrgh the energy relation for ideal flow postulates the velocity maximum shall be at the suction
race of the channel. the secondary flow induced by turbulence and othcr effects from the pressurc face
towards the suction face shifts the velocity maximum in the main flow
at
the outlet of
a
rotor channel
towards the pressure face. Sirnultancously the accumulation of bot~ndar? layer matcrial on thc
suction face towards the outlet of the channel tends to stall there at the rotor exit, rotating with blade
speed.
Will1 increasing streamwise channel length, this erect counteracts the energy conversion of a runner
(impeller) channel. One remedy consists
in
the choice of a short charinel together with an adcquate
number of rotor vanes. For the usual runner channel of turbines with forivard curved vanes, the
cffccts of curvature and Coriolis force are additive. For backward curved vilnes, as are employed in
purnps, pump-turbinzs and psrtially also in turbines, both effects act against each other. Howe\er,
a balance is not possible,
as
it
would annul the rotor's work.
As
was demonstrated by
Sirscheletzky,
also an ideal flow may have srailed zones in consequcnce of
the principle of least action.
In the rotor channel, the width of such a stalled zone could also ba pre-
dicted by
mans of this principle
[5.59].
A
criterion for the lninilnum vane number, t\hicli prcvcnt
stall, could be derived from this by the author
[5.59].
5.5.5.2.
Interaction
of
the
boundary layer and the
main
flow
Whereas the afcrementioned secondary flow may be spread over the main and boundary
layer
flow,
the followi~lg flow superimposed on the first, is usually due to the boundary
layer. This follows from the interaction of
a
pressure gradient across the main flow and
the boundary layers of
the
walls, covering the main flow in the direction, in which the
retained pressure acts.
Owing to the relatively slow flow in the boundary layer, the pressure gradient from the
main flow due to curvature and Coriolis force
is
not balanced by these velocity-linked
forces
withiil the layer. Therefore the fluid in the boundary layer is displaced from the
pressure to the suction face of
the duct.
[Jsually this shift is balanced by a niotion of the main flow in the mean stream facc of the
duct. Thus the secondary flow due to this effect usually consists of two loops turning
against each other.
This may occur in a rotor channel (Fig. 5.5.2b) or in a Send
{Fig.
5.5.2~).
The return flow in the mean stream face transports fluid of the main flow from the sucticn
to
the
pressure face, thus distorting the original velocity distribution across the channel
as
the fluid flows downstream by shifting the velocity maximum from the pressure to the
Suction face. The consequence is essentially the same
as
that of turbulence-conditioned
cross motion, described before.
-
$1
1.h~
erfccts
of
ro~ation on boundary
layer
in
a
turbomachine rotot.
Rep.
MD-24,
Mech.
Engg.
Dept.
Stanford
University
1970.