J . Fluid Mech. (1966), vol. 25, part 2, p p .

401-413 401
Printed i n Great Britak

The mechanics of the formation region of vortices

behind bluff bodies
By J. H. GERRARDt
Department of the Mechanics of Fluids, Manchester University

(Received 22 September 1965)

The characteristic lengths of the oscillating wakes of bluff bodies is discussed;

in particular, those used in the universal non-dimensional frequencies proposed
by Roshko (1954b) and Goldburg, Washburn & Florsheim (1965). It is concluded
that these are equivalent at high Reynolds number. A closer examination leads
to the conclusion that there are two simultaneous characteristic lengths; the
scale of the formation region, and the width to which the free shear layers diffuse.
Discussion of the mechanics of the formation region results in a physical basis
for the determination of the frequency by these two characteristic lengths.
The ideas developed are applied to the effects of splitter plates in the wake. The
possibility of a high-Reynolds-number symmetrical formation region is suggested
as an explanation of the very small lift values observed in the absence of free-
stream disturbances.

1. Wake width as characteristic length

I n a recent publication Goldburg et al. (1965) show that the ‘total wake mo-
mentum thickness,’ 8, is the relevant characteristic length to be used in the
formation of a universal non-dimensional frequency, S, = N e / U ,where N is the
fundamental wake frequency and U is the free-stream speed. They make no
reference to the work of Roshko (1954b)’ in which he defines a similar universal
number S* = Nd’/U, where d’ is the wake width determined by his ( 1 9 5 4 ~ )
notched-hodograph theory, and U, is a wake velocity simply related to the base
pressure coefficient C,. The coefficient of drag per unit length, C,, is similarly
related to C, by the notched hodograph theory. Goldburg et al. show that for
two-dimensional flow 8/d = +I?,, where d is the length used in forming the drag
coefficient.
Roshko achieves good collapsing of the frequency as a function of Reynolds-
number curves for the circular cylinder, normal flat plate and 90” wedge at the
higher Reynolds numbers. There is some separation of the curves for two of the
bodies at a reduced Reynolds number, R*,of 4000. When, however, results for a
circular cylinder from Gerrard (1965) are compared with Roshko’s S*, R* plot,
considerable discrepancy is found. This is because the Strouhal number (NcZ/U)
is the same in both experiments, but the values of d‘ and Us are quite different,
in the low free-stream turbulence measurements, from the values Roshko
t During 1965, National Science Foundation Senior Foreign Scientist Research Fellow
at Pennsylvania State University, Department of Aeronautical Engineering.
26 Fluid Mech. 25
402 J . H . Gerrard
determines. At R* = 4000, Roshko’s values of S* lie between 0.16 and 0.20,
Gerrard’s value of S* at this reduqed Reynolds number is 0.29.
The results of Roshko and Gerrard have also been examined on the basis of a
plot of SO as a function of Re following Goldburg et al. There is again considerable
spread of the results at values of Re of the order of 1000. The spread is much less
at higher Reynolds numbers though not as small as the result Roshko publishes.
There seems little to choose between the basic ideas of Roshko and of Goldburg
et ab. as the Reynolds number approaches
__ the critical. The ratio of the non-
dimensional frequencies

When this is computed for a circular cylinder it is found to be a quite rapidly

varying function of Reynolds number below a Reynolds number of 30,000.
Values of C, were taken from Goldstein (1938), Us/U from Gerrard (1965)
and d’/d from Roshko (1954b). At a Reynolds number of 3000 the ratio is 1.9
or greater. The ratio steadily decreases with increasing Reynolds number to
about 1-3 at R = 30,000. Reasonable extrapolation suggests that the ratio does
not fall below 1.2 by the time the critical Reynolds number is reached. It appears
that Roshko’s suggestion is correct, that a single parameter (R,or R,) in-
sufficiently describes the phenomenon when the position of transition is moving
towards the body in the shear layers which later roll up into vortices. Unfortu-
nately, this transition region extends from a Reynolds number of about 400 to one
of about 40,000. Examination of Roshko’s (1954b) figure 5 strongly suggests that
when the flow is entirely laminar the Strouhal number itself is the required
universal non-dimensional frequency. At high Reynolds number, Roshko’s
and Goldburg’s methods are equally good in producing a universal non-dimen-
sional frequency. Both produce discrepancies as the relative proportion of
laminar shear layer in the formation region increases.

2. Two simultaneous characteristic lengths

Many of the flow parameters are dependent on the turbulence level of the free
stream: as the author (1965) has shown, this is reflected in a dependence on cylin-
der diameter as well as on the Reynolds number. There are only very small
concomitant changes in the Strouhal number. Figure 1, obtained at lowturbulence
level, shows that the average Strouhal number is no different from that obtained
at higher turbulence level and that the exact form of the relationship varies from
day to day, presumably with the disturbance level in the wind tunnel. The con-
stancy of the Strouhal number is as yet unexplained. On Roshko’s theory, refer-
ring to the values of Us determined by the author (1965), we would expect just
as large a change in the Strouhal number as we observe in Us: this change would
be amplified by the attendant change in d‘.
Roshko (1954b) considers average values of d’ and Us,which in his experiments
varied little over the whole Reynolds-number range. It is instructive to apply
Roshko’s method in detail: the experimental results of Roshko (1954b) and
Gerrard (1965) are plotted in figure 2. Clearly S* is not a universal non-dimen-
sional frequency.
 Formation region of vortices behind bZuff bodies 403
When there is no interference, in the form of splitter plates, in the wake, one
may expect that the width d' and the length 1, of the formation region will be
equally good characteristic lengths. When splitter plates are inserted in the
wake we shall see below that I , appears, in some cases at least, to be more signifi-
cant. The use of an experimental quantity rather than d' is also an attraction.

0 21 Y

O 103 2
019 O E
104
R

0.3

0.25

8,

02 ~

015
404 J . H . Gerrard
We shall consider, therefore, a non-dimensional frequency N1,lU or Sl,/d.
Little qualitative difference results from the use of U rather than U,.The velocity
chosen should be representative of the shear layers before they roll up into vor-
tices. Such a speed is expected to lie between U and [Link] non-dimensional
0.6

0.4
2
4

0.2
102 103 104 105
R
FIUURE 3. The first characteristic length, If the length of the formation region. S' = Nl,lU
is plotted as a function of Reynolds number. It is assumed that the observed (Bloor 1964)
dependence on cylinder diameter reflects a dependence on free stream turbulence.

102 103 104 105

R
4. The second characteristic length, L,the diffusion length, plotted as a function
FIGURE
of Reynolds number. L is defined such that S'LJd = 0.2.

frequency S' = Nl,/U is plotted in figure 3. The assumption is made that Bloor's
(1964) small cylinder-diameter values of I , apply to the higher free-stream turbu-
lence case.
We shall see in $ 6 that this assumption may be incorrect and that, as with
d', we expect the maximum I , in the turbulent stream to be significantly less
than the maximum I , in the undisturbed flow. This has no qualitative effect
on the arguments proposed before $6.
 Formation region of vortices behind bluff bodies 405
We now introduce a second characteristic length L, such that S'Lld is indepen-
dent of Reynolds number. The values of Lld corresponding to S'LId = 0.2 are
plotted in figure [Link] value 0.2 is arbitrarily chosen, but the resulting value of L
agrees in order of magnitude with the physical interpretation which will be
suggested below.
After a physical discussion of the mechanics of the fluid flow in the formation
region, we shall return to the problem of seeking opposing tendencies which
result in the constancy of the Strouhal number. Insufficient data are available
to take the discussion beyond a consideration of the circular cylinder.

3. The size of the formation region

The author (1965) has shown the length of the formation region of the vortices
behind a circular cylinder to be a relevant length scale for the distribution
of fluctuating velocity with distance close to the body. The variation of length of
the formation region and of Roshko's (19543) wake width d', with Reynolds
number, is similar when the free-stream turbulence level is low.
Little has been published on the region of flow close behind a bluff body,
which it is recognized, plays an important role in the determination of the
frequency of vortex shedding. A physical discussion of the mechanics of the

FIGURE
5. Filament-line sketch of the formation region. Arrows showing reverse flow
(c) and entrainment (a) and ( b ) .

formation region will serve to introduce a relationship between the size of the
formation region, the strength of the vortices and the frequency of vortex shed-
ding. The most dramatic changes take place in the formation region as the
Reynolds number is increased from about lo3to about 5 x lo4 and the transition
to turbulence moves upstream in the shear layers (Schiller & Linke 1933, and
Bloor 1964). During this movement, the formation region shrinks in size for a
reason which we can make clear by reference to figure 5.
Bloor (1964) defines the end of the formation region as the point on the wake
axis, closest t o the cylihder at which a hot-wire signal, characteristic of the
oscillating wake downstream, is observed. This implies that the end of the forma-
tion region is where fluid from outside the wake first crosses the axis. Fluid is
drawn across the wake by the action of the growing vortex on the other side.
Figure 5 illustrates the formation region by means of filament lines within the
rolling-up shear layers. It is shown a t the instant when irrotational flow is begin-
ning to cross the wake axis. The arrows show the path of this fluid at later times:
it is partly entrained by the growing vortex and partly by the shear layer up-
stream of the vortex. Some of this fluid will also find its way into the interior of
406 J . H . Cerrard
the formation region. The fluid must bear vorticity of opposite sign to that of the
entraining layer because the irrotational flow entering the wake cannot cross
the filament line springing from the lower boundary-layer separation point in
figure 5.
The entrainment of fluid bearing vorticity from the other shear layer takes
place periodically, but entrainment by the shear layer is a continuous process.
The vorticity in the interior of the formation region, apart from the periodic
flow described above, may be expected to be considerably less than that of the
shear layers. I n this case, the periodic entrainment of fluid across the wake will
have a predominant influence.
The size of the formation region is determined by the balance between en-
trainment into the shear layer and the replenishing of fluid by the induced
reversed flow described above. Changes in entrainment with Reynolds number
will be governed by the way in which the flows, marked a, b and c in figure 5,
separately vary. We can argue a posteriori that the circulation carried by
flow a does not vary significantly with Reynolds number. The entrainment flow
b will be governed mainly by the length of the turbulent shear layer. As the
circulation carried by flow b increases with Reynolds number, we must assume
that that carried by flow c decreases. Much of the circulation of flow c will remain
in the interior to be effectively cancelled half a period later. Some cancellation
will occur at the rear surface of the body.
If from the equilibrium situation, at one Reynolds number, the Reynolds
number is increased, the position of transition will move towards the cylinder
and the rate of entrainment will tend t o increase. The reversed flow is unable
to increase because the forming vortex will tend to be weaker the greater the
entrainment, since fluid bearing oppositely signed vorticity will, on the whole,
be entrained. Hence the formation region must shrink in length.
The validity of this argument is substantiated by the following simple treat-
ment. By integrating the turbulent shear-layer velocity distribution given
by Schlichting (1955), we find that the volume flux increases like the mean speed
multiplied by the distance from the origin of the layer (at large distances from
the origin). If we neglect the laminar entrainment compared with that of the
turbulent part of the layer, we may put the rate of entrainment proportional
to U,l,, where Us is the speed outside the boundary layer at separation, and I ,
is the length of turbulent shear layer in the formation region. This volume
flux must balance the reversed flow if the formation region is to remain the same
size each period. At low Reynolds numbers, where the vortices are laminar,
let us make the assumption that the entrainment flow b is negligible compared
with that under turbulent conditions found at higher Reynolds number. If
the entrainment bears fluid of opposite vorticity to that in the shear layer, we
expect the reduction in the strength of the vortices below the laminar value to
be proportional to the difference in turbulent entrainment. That the strength
of laminar vortices is considerably less than the circulation shed in one period
from the separation point, implies that the circulation carried by flow a is quite
large. We have assumed that the flow a circulation (suitably non-dimensionalized)
does not vary with the Reynolds number.
 Formation region of vortices behind bluff bodies 407
Using vortex strengths determined by Bloor & Gerrard (to be published),
and the lengths of turbulent ahear layer determined by Bloor (1964) we find
that the proportionality does approximately exist between the rate of tur-
bulent entrainment and the change in vortex strength. This is shown in
figure 6 . The differences in vortex strength are plotted non-dimensionally as the
strength of laminar vortices, taken to be 0.8, minus the strength of the turbulent
vortices at the particular Reynolds number. The vortex strength K is non-
dimensionalized by division by m u d . The bifurcation of the entrainment curve

022 -

3 020

018 -
Low
\ turbulence

102 103 104 105

R
FIGURE
6. Strouhal number S = N d / U , entrainment volume flow (lJ,/U)(lt/d),and laminar
vortex strength minus turbulent vortex strength, as a function of Reynolds number. The
high-turbulence curves are derived from Roshko ( 1954b) the low-turbulence curves from
Gerrard (1965).

corresponds to the dependence upon diameter as well as Reynolds number

referred to above. Curves corresponding to high and low turbulence level are
also shown in the vortex strengths. The high-turbulence vortex strengths are
determined from the results of Roshko (1954b), as will be explained in the publica-
tion by Bloor & Gerrard. Thus it is shown that the changes in the strength of the
vortices with Reynolds number can be attributed to the variation of entrain-
ment of vorticity-bearing fluid by the turbulent part of the sheax layer.
408 J . H . Gerrard

4. The frequency of vortex shedding

The growing vortex continues to be fed by circulation from the shear layer
until the vortex becomes strong enough to draw the other shear layer across the
wake. The approach of oppositely-signed vorticity in sufficient concentration
cuts off further supply of circulation to the vortex, which then ceases to increase
in strength. We may speak of the vortex as being shed from the body at this stage.
We postulate that this is the basic mechanism determining the frequency of
vortex shedding. Roshko (1954b) has shown that the frequency increases if the
scale of the formation region is reduced. When the shear layers are brought
closer together their interaction is facilitated and the periodic time shortened.
We have already shown that, though the above may be described as the basic
mechanism, there is another major effect. This we suggest to be the thickness
of the shear layer when it reaches the region of strong interaction at the end of the
formation region. The significance of this was implied in Fage & Johansen’s
paper in 1927, where the mean thickness of the shear layers springing from
different bodies was shown. The importance of the shear-layer thickness was
explicitly mentioned by Berger in 1964. We do not differentiate between a thick
shear layer spread under the action of turbulent diffusion and an effectively
thick shear layer formed by its rolling up into a succession of concentrated
vortices. An increase in the turbulence in the shear layers will result in their
being more diffuse in the region of interaction. When the layer is diffused it will
take longer for a sufficient concentration of vorticity to be carried across the
wake and initiate shedding. So we expect the shedding frequency to decrease
as the ‘diffusion length ’ L increases. On the same reasoning, the greater diffusion
of the vorticity will result in less entrainment into the growing vortex, hence a
higher value of the vortex strength a t high Reynolds number. I n this case there
will be an effective cancellation of circulation in the interior of the formation
region.
The relative constancy of the Strouhal number over the whole Reynolds-
number range from about 400 to the critical Reynolds number follows from the
fact that the two major frequency-determining factors tend to change the
frequency in opposite directions as the Reynolds number is altered (compare
figures 3 and 4). When the free-stream turbulence level is increased from a low
value a t constant Reynolds number, the length of the formation region decreases,
but the diffusion length presumably increases and to such an extent that the
opposing frequency changes balance. That they balance is not fortuitous for
the increased entrainment attending the increase in turbulence level is respon-
sible for both the decrease in size of the formation region and for the increase
in the diffusion length. The width of a turbulent shear layer and the entrainment
into it are both proportional to the length of the layer.
Two major factors which determine the frequency of vortex shedding have
been postulated. Kronauer (1964) has proposed the theory that the frequency
is mainly determined by the feedback of velocity fluctuations to the boundary-
layer separation point from the wake. It is the author’s contention that, while
this feedback is essential to the production of the oscillating wake, it is not the
 Formation region of vortices behind b l u 8 bodies 409
primary factor in the determination of the frequency. Work by the author, as
yet unpublished except in outline (Gerrard 1963), suggests that without this
feedback, which causes the separation points to oscillate, the shear layers will
roll up without significant interaction: the shear layer will escape entrainment
and the frequency-determining mechanism will have no opportunity to work.
Independent roll-up of the shear layers is characteristic of the observed wake
development between Reynolds numbers of about 40 and 90.
At higher Reynolds numbers, the turbulent character of the frequency-
determining mechanism explains why the frequency is not well defined. The
spectra of oscillating quantities are narrow bands of noise rather than pure tones.

5. The effect of splitter plates in the wake

To be acceptable, any theory of the frequency-determining mechanism
must be in accord with the observed effects of splitter plates in the wake of bluff
bodies. I n 5 2 the constancy of the Strouhal number over a wide range of Reynolds
number is attributed to the constancy of the product ZfL. Thus we expect the
Strouhal number to increase if 1, or L decrease.

0 1*o 2.0
W
FIGURE
7. Strouhal number as a function of plate length with a
splitter plate on the wake axis.

Some experiments were made at a Reynolds number of 2 x lo4 on a 1 in. dia-

meter cylinder with a spvtter plate attached as shown in figure 7. The frequency
of vortex shedding was determined with a hot-wire anemometer in the wake
and a wave-analyzer attached to the amplifier output. At this Reynolds number
the length of the formation region was expected to be about one diameter
(that is, to terminate one radius downstream of the rear of the body) since the
Pennsylvania State University low-speed wind tunnel did not have a low-turbu-
lence level.
410 J. H . Gerrard
Figure 7 shows that as the length of the splitter plate is increased up to and
even beyond the length of the formation region, the Strouhal number is decreased.
This decrease is expected if the length of the formation region is increased by
the insertion of the plate. The diffusion length will also increase if 1, increases.
Physically, we expect the cross-flow resulting in the shedding of the eddy to be
less easily set up in the presence of the splitter plate, some of the cross-flow
will be deflected away from the other side of the wake; this results in an increased
period. The production of circulation a t the rear of the cylinder is expected to
increase when a splitter plate is present. This results in the weakening of the
vortex strength in the early stages of its growth which would also result in a
decreased frequency. This effect is expected to be small.
This experiment shows the fundamental importance of the flow in the interior
of the formation region in determining the frequency.
Experiments were also made repeating the work of Roshko (1954b) using
splitter plates with a gap between the cylinder and the plate. The frequency
measurements were made at a cylinder Reynolds number of 2 x lo4 again,
which was close to that used by Roshko. When the gap length was small, the
reduction in Strouhal number (for the smaller plate lengths) was the same as
that found with no gap. For larger gaps there is a large effect of the gap or the
leading edge of the plate. It was found that the relevant parameter was the width
of the gap rather than the position of the trailing edge of the plate.
Flow visualization showed that the spectacular discontinuity in the Strouhal
number observed by Roshko (1954b) was a transition between a flow regime
in which the formation region was lengthened so as to include the plate in its
interior, and a flow r6gime in which vortices formed upstream of the plate. The
discontinuities were found to occur at the same gap length; the position of the
trailing edge of the plate had only a minor effect for plate lengths of 0.7 to 1-14
diameter. The greater reduction of frequency produced by the Roshko-type
plate with a gap may be explained: if some vorticity crosses to the other side of
the wake upstream of the plate in insufficient quantity to induce shedding, it
will be longer before the weakened vortex will eventually be shed downstream
of the plate. As soon as the cross-flow upstream of the plate is capable of producing
shedding, the flow will undergo transition to the higher-frequency mode.
A series of measurements of frequency were made with one plate 0-69diameters
in length placed normal to the free stream behind the circular cylinder at the
same Reynolds number. Again there was a large discontinuity in Strouhal
number as the distance of the plate from the cylinder was varied; as shown in
figure 8. Though the jump is referred to as a discontinuity in both this experiment
and the one described above, there was a range of position close to the jump in
which the flow intermittently displayed both frequencies. Again the formation
region expanded to include the plate when it was close to the body. At first
sight this appears to contradict our expectation that a larger formation region
is expected to generate a lower frequency. As far as the shedding of the growing
vortex is concerned, however, the effective formation-region length is the dis-
tance from the plate to the end of the region. We conclude that the reduction
in this length is responsible for the increase in Strouhal number. A physical
 Formation region of vortices behind bluff bodies 41 1
explanation of the mechanism is also possible. When a vortex is growing close
behind the plate, there is a l%rgecross-flow velocity produced near the plate;
this will facilitate the shedding process and increase the frequency. The channel-
ling of the flow between the plate and the vortex will also reduce the effective
diffusion length L of the layer. Reference must be made to the experiments of
Bearman (1965), in which a splitter plate parallel to the free-stream direction
produced an increase in Strouhal number. I n his work, the boundary layers
were turbulent on separation and the total boundary-layer thickness at the trail-

0 1.0 2.0 3.0

Jld
FIQURE
8 . Strouhal number as a function of plate-cylinder separation with a
splitter plate normal t o the wake axis.

ing edge of the body was one half of the base height. It appears that entrainment
by the separated layers reduces the wake width by a tendency toward reattach-
ment of the boundary layer downstream of the step. The length of the formation
region is reported to increase, however. This experimental result only fits in
with Roshko’s (1954b ) finding that the more bluff the body, the lower the fre-
quency. It seems likely that this different configuration, in which the shear layers
are not considerably thinner than their separation, has different properties from
the wide wakes we have considered above.

6. The effect of free-stream turbulence on vortex formation

An explanation has been offered of the relative constancy of the Strouhal
number over the whole range of Reynolds number in which turbulent vortices
are observed. Similarly, on the basis of the two characteristic lengths determining
the frequency, a physical explanation of the effects of splitter plates has been
given. It appears that one may explain the effect of increasing the free-stream
412 J . H . Gerrard
turbulence from very small values in the same way, as we did in $ 2. On increasing
the turbulence level, the scale of the formation region shrinks and one may
hypothesize that the diffusion length'will increase. Opposing tendencies of fre-
quency variation could cancel out, and hence the Strouhal number be little
affected. This explanation has an unsatisfying facet.
The author (1965) has shown that when the disturbance level is very low, the
fluctuating lift and fluctuating velocity at the shoulder of the cylinder change
by about two orders of magnitude between Reynolds numbers of 2000 and
lo5. The variation of the vortex strength and the Strouhal number is less than
50 %. It seems unlikely that these facts can be explained simply by a change in
scale of the formation region. A change in length of the formation region by a
factor of two would be expected to change the amplitude of the lift by less than
a factor of four. Further, a change in the free-stream turbulence does not alter
the range of variation of the length of the formation region but only the Reynolds
number at which it occurs. This statement depends upon our assumption that
Bloor's (1964) values of I,, obtained with small diameter cylinders, correspond
to the high-turbulence case. Roshko's (1954) wake width d', on the other hand,
attains higher values in the range, 700 < R < 7000, when the disturbance level
is low, than its maximum value at higher-turbulence levels. The lift also only
drops to low values when disturbances are absent. Thus we are lead to suggest
a possible fundamental change in the mode of vortex formation. There is a
possibility that, as suggested in $4, in the absence of free-stream disturbances,
the vortex sheets are able to develop independently of each other. This implies
a transition to a type of flow similar to, if not the same as, that obtained a t very
low Reynolds numbers. The shear layers in this case will be turbulent as Bloor
(1964) has shown. Though it must be pointed out that it has not been shown
that lift or velocity fluctuation at the shoulder of the cylinder fall to low values
when there is a probe in the shear layer downstream.
Laminar vortices, it is believed, are usually formed in the manner described
in $ 4 down to a Reynolds number of about 90; and so a reversion simply to
laminar shear layers would not produce an explanation of the observed fall in
lift which seems to require a symmetrical configuration, whether laminar or
turbulent. The possibility of a symmetrical configuration, with two standing
vortices, is present only a t a Reynolds number of a few thousand, because it is
only in this range of Reynolds number that the formation region is large. I n
this condition, the removal of disturbances will allow independent development
of the two shear layers. If the explanation of the extremely low lift forces observed
is to be found in the idea of a symmetrical formation region, another difficulty
appears. It has been considered that the wake oscillation observed at low
Reynolds number with symmetrical flow near the body is due to the instability
of the narrow wake. If at high Reynolds numbers the same is true, why is the
Strouhal number the same with symmetrical and asymmetrical formation
regions? Why, indeed, is the Strouhal number only just detectably different
a t the low Reynolds-number transition ? One also wonders whether three-dimen-
sional effects would need to be absent to allow the formation of the symmetrical
pattern.
 Formation region of vortices behind b l u , bodies 413

7. Conclusion
We conclude that the entrainment of fluid from the interior of the formation
region and its replenishment by reversed flow is fundamental to the determina-
tion of a scale which determines the frequency of vortex shedding. A second
characteristic length of fundamental importance to the determination of vortex-
shedding frequency is, what we have termed, the diffusion length. This is the
thickness of the shear layer at the end of the formation region where the layer
is drawn across the wake. The cross-flow presents fluid for entrainment which
bears vorticity of opposite sign to that of the shear layer which is entraining
the major portion of the fluid a t that instant. The physical basis for this choice
of two simultaneous characteristic lengths of the shedding process is discussed.
On the basis of this mechanism the effects of splitter plates on the frequency of
vortex shedding have been explained in the case of bluff bodies with wide wakes.
It is suggested that there is a possible high Reynolds-number flow regime in
which the formation region is symmetrical in the absence of free-stream dis-
turbances, This could account for the diminutive lift values observed under
these conditions (Gerrard 1965) .
The author cannot adequately acknowledge all who have helped to germinate
the ideas which have resulted in the present paper but would like to appre-
ciatively mention E. Berger, M. S. Bloor, A. G. Fabula, M. B. Glauert, N. H.
Johannesen, R. E. Kronauer, M. J. Lighthill, J. L. Lumley, M. V. Morkovin,
P. D. Richardson and A. Roshko, and to apologise to those whom he has omitted.
Thanks are also due to the National Science Foundation for the Senior Foreign
Scientist Research Fellowship which the author held when completing this
work.
REFERENCES
BEARMAN, P. W. 1965 Investigation of the flow behind a two dimensional model with
blunt trailing edge and fitted with splitter plates. J . Fluid Mech. 21, 241.
BERGER, E. 1946 Jahr. W i s . Ges. L. & R., Berlin.
BLOOR, M. S. 1964 The transition to turbulence in the wake of a circular cylinder.
J . Fluid Mech. 19, 290.
FAGE,A.& JOHANSEN,F. C. 1927 The structure of vortex sheets. Aero. Res. Counc. R & M
no. 1143.
GERRARD,J. H. 1963 The calculation of the fluctuating lift on a circular cylinder and its
application to the determination of aeolian tone intensity. AGARD Rep. no. 463.
GERRARD,J. H. 1965 A disturbance sensitive Reynolds number range of the flow past
a circular cylinder. J . Fluid Mech. 22, 187.
GOLDBURG,A., WASHBURN, W. K. & FLORSHEIM, B. H. 1965 Strouhal numbers for the
hypersonic wakes of spheres and cones. A . I . A . A . J . 3. 1332.
GOLDSTEIN,S. 1938 Modern Developments on Fluid Dynamics, vol. 2, figure 149. Oxford
University Press.
KRONATJER, R. E. 1964UnpubIishedpresentationatI.U.T.A.M. [Link]
vortex motions, Ann Arbor.
ROSHPO, A. 1954a A new hodograph for free-streamlinetheory. N A C A T N no. 3168.
ROSHKO, A. 1954b On the drag and shedding frequency of two-dimensional bluff bodies.
N A C A T N no. 3169.
SCHILLER, L. & LINKE,W. 1933 2. Flugtech. Motorluft. 24, 193.
SCHLICHTING, H. 1955 Boundary Layer Theory. London: Pergamon Press.