Sunteți pe pagina 1din 7

REVIEW: Turbulent Developing

A.Klein Pipe FlOW


Motoren- und Turbinen-Union,
MiinchenGmbH, D8000 Experimental results on turbulent developing pipe flow are reviewed. Upstream
Munchen, Germany conditions are shown to have a large effect, in particular on the development of the
velocity profile. It is demonstrated also that the blockage factor, when plotted
against the Reynolds number defined with flow distance, is the parameter best
suited to indicate the onset and termination of transition.

Introduction
Until recently, flow development in pipes was believed to be where
solely determined by a flat-plate like growth of the wall-
boundary layers. Therefore, the peakiness of the velocity > = V(i-JL)JLd(JL\ (2)
profile, defined as the ratio between the maximum value and R Jo V UJ R \ R /
the spatial mean, was thought to increase gradually until the Some researchers used a two-dimensionally defined
boundary layers meet at the pipe centerline. According to displacement thickness
measurements quoted by Schlichting [1] this appeared to
occur in turbulent flow after 25 to 100 pipe diameters, the
exact length depending on the Reynolds number. Thereafter
the flow was considered fully developed, that is, the flow
'-T-lK'-iX-r)-
In these cases the blockage factor was calculated from
properties were thought to have become independent of inlet 2a
itwo _ / SltwoN 2 _
conditions and to remain unchanged as the flow proceeded B = ( 4 )

further down the pipe. In fact, however, developing pipe flow


is much more complex. Bradley (quoted in [2]) appears to be In Fig. 1 the blockage factor has been plotted against
the first who discovered that velocity profile peakiness may normalized pipe length for all those cases where these
reach a maximum, which he found at about 40 pipe distributions exhibit a maximum. The entry conditions of the
diameters, and then decrease again. Weir, Priest, and Sharan various experiments, as far as they are known, also are given.
[3] also obtained a maximum core velocity at 40 pipe In addition to Reynolds number ReD = u D/v and centerline
diameters, while gradual velocity increases along the pipe turbulence intensity Tuc, the caption to the figure includes
were measured by Deissler [4] and Sale [5]. The latter type of such details as the area ratio of the contraction upstream.
velocity distribution was produced in [3] also, when a trip ring It can be seen immediately that all the maxima of the
was inserted at the pipe entrance (see also Sharan [6]). Thus it distributions are located at x/D = 35 to 40. The results of
was shown that a change of inlet conditions may create Bradley (see [2]) and Miller [9] (for the lower of the two
substantial differences in flow development. To investigate Reynolds numbers) very well confirm those of Barbin and
this observation further, all the experimental results on Jones [10], which are usually considered "classic", since a
turbulent velocity-profile development in pipes known to the false datum (nonzero displacement thickness at inlet) and
author are reviewed. Since the performance of diffusers disturbances in the wall region were avoided by a boundary-
depends on the flow conditions at the ends of their entrance layer bleed, while laminar-turbulent transition phenomena
pipes, the present work is a necessary first step towards the were excluded by use of a small contraction ratio and a sand-
comparison of dif fuser data presented in the same issue of this grain trip. Larger Reynolds numbers are seen to produce
Journal [7]. lower peak values and smaller Reynolds numbers higher peak
values:max =0.155 for ReD = 13-105 (Miller [9]) and 0.215
Turbulent Pipe Flow for ReD = 1.8-105 (Cockrell and Markland [11]). Beyond
their maxima the curves show different amounts and locations
The measure of flow development used in this review is of "undershoot" compared with the fully developed values.
Sovran and Klomp's [8] blockage factor which is defined by This waviness is due to the adjustment of the shear stress in
the ratio of the axisymmetric displacement thickness 5iax to the central region, an extremely complicated process since
pipe radius R or of the ratio of the centerline velocity uc to the shear stresses of any direction exist (see Bradshaw [12]).
spatial mean velocity u (the bulk velocity) Nevertheless, there is good agreement between the Refl =
4-105 curve from [3] (full triangles in Fig. 1) and the single
data points due to Bradley and Cockrell [13] and to Patel [14].
Figure 1 also contains the classical fully developed flow
data of Nikuradse [15]1 for four Reynolds numbers. They are
Contributed by the Fluids Engineering Division for publication in the
JOURNAL OF FLUIDS ENGINEERING. Manuscript received at ASME Headquarters Note that Schlichting's [1] power law presentation of the Nikuradse data
February 19, 1981. results in somewhat higher blockage factors.

Journal of Fluids Engineering JUNE 1981, Vol. 103/243


Copyright 1981 by ASME
Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 08/12/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
shown as asymptotic values for very long pipes, although they From Fig. 1 one may deduce that his data are indeed
were obtained with x/D = 100 (for Re D = 1.0-105), x/D = representative of fully developed conditions.
120 (for Re D = 2.0-10 5 and 4.0.10 5 ) and x/D = 70.5 (for Figure 1 exhibits one more interesting feature. It reproduces
Re D = il10 5 ). The last pipe length appears to be rather two curves from [3], obtained at approximately the same
short for full flow development to occur, but Nikuradse Reynolds number in the same test facility with the only
reports that tests with x/D = 40, 65 and 100, conducted at distinguishing feature that the blockage factors were
Re D = 9.0* 10 5 , showed exactly the same velocity profile. calculated from 5,ax/Z) in one case (full triangles) and from

0,22

AC\ \ 5

A
B [14l, KE[
L/N
0,18
\
// 4k, 1 ' \
0,16 X-l ? > ^=J^22 _ A = A As
0,14 't\ JB Qj rt
T
j n -n
"^^"*
0,12 ///
0,10
J Ilk 1

4i
\J1
1
0,08
Nikuradse t15jy / 7
V
ReD= 1.0-10 5 -' / /
0,06
ReD=2,0-105-///7
1 [l3],ReD=2,2-105'
0,04- JI S ReD = 4,0 -10 s ' /

0,02-
I ' ReD=11 -10 s '


0
10 20 30 40 50 60 70 80 90 100 110 120 X /Q 140

Fig. 1 Development of the velocity profile w i t h n o n d i m e n s i o n a l


distance from inlet: cases where the blockage-factor distribution has a
maximum

Sym- Ref. 10 5X Tu c [%] Contr. Obtain-


b o l No. Re
D at entry r a t i o ed from Remarks

A 4.3 0.25 ^Lax n a t u r a l development


A 3 2 to 3 16:1 D w i r e gauze a t e n t r y
A 4.0 0.25 uc/S n a t u r a l development

+ 11 1.8 7 7

D 5.5 1.0 quarter-circle nozzle


9
13.0 1.0 12:1 entry

m 2 7 7 7

1.6 7 7
4 u square-edged entry
* 5.8

10 3.9 0.5 4:1


b . - l . - b l e e d plus
sand g r a i n t r i p

X 17 6.0 6:1
L /D = 1 5 , s c r e e n r i n g
7
h/D = 0 . 1 0 h i g h

t 16
2.5
5.0
0.4
0.3
16:1 ^lax
D
turbulent
tripping
device

B 13 2.2 7 7 uc/=
uc/tl
14 2.7 7 4:1

Nomenclature

ks
= height of equivalent sand u = velocity, uc value at pipe
B = blockage factor (see roughness centerline
equation (1)) Re D ,Re T = Reynolds number, based u bulk velocity
D = 2 R = pipe diameter on ii and on D or x, x coordinate along pipe
H{2 = boundary-layer shape respectively axis, x = 0 at pipe entry
factor, ratio between the r = coordinate in the radial axisymmetric boundary-
axisymmetric boundary- direction layer displacement thick-
layer displacement and Tu = turbulence intensity in ness (see equation (2))
momentum thickness the axial direction, Tu,. kinematic viscosity of
h = height of trip ring value at pipe centerline fluid

244/Vol. 103, JUNE 1981 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 08/12/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


120 X/D 140

Fig. 2 Development of the velocity profile with nondimensional


distance from inlet: cases where the blockage-factor distribution has
no maximum

Sym- Ref. 10 5X ^c [*] Contr. Obtain-


bol No. ratio ed from Remarks
at entry

6 1.0 0.25 16:1


<flax
trip ring at entry,
h/D = 0.062 high
4.0 D

3 4 1.6
5.8 7 ?
bellmouth entry,
blow-down facility

5 4.0 1.0 20:1 5 -diffuser upstr,


of contraction

13 2.2 ? ?
>-ir rough, sand paper
ks/D = 40-10 -4

19 7.8 25:1 rough k_/D - 3.8-10^


?
% 25.0 pipe kg/D = 2.0-10~
4

u/uc in the other case (open triangles). The discrepancy is configuration was altered, but it never vanished. As in the
considerable, although both procedures should of course tests employing the square chamber, disturbances were
produce identical results. It may be due to marked asym- confined entirely to the boundary layer.
metries of the turbulence distribution at the inlet which were These experiments clearly demonstrate the extreme sen-
noted in these experiments, although the velocity distribution sitivity of developing pipe flow to upstream conditions. They
at the inlet was symmetrical (see [6]). But a more general also show that a truly axisymmetric pipe flow does not exist.
explanation can be deduced from an investigation recently Hence determination of the blockage from u/uc (where u is
conducted by Pozzorini [16]. Air was sucked through filter derived from the flow rate) will generally be more accurate
mats into a chamber of square cross section, through a than from 8iax/D, especially if the latter is calculated from
bellmouth, a settling length with turbulence screens, and measurements in a single plane only.
through a carefully shaped 16:1-contraction with a trip close The cases where profile peakiness does not exhibit a
to its outlet optimized so that the boundary layer was com- maximum as the flow proceeds along the pipe, depicted in
pletely turbulent around the entire circumference. The Fig. 2, show the same amount of diversification as the curves
peripheral total-head distribution within the boundary layer in Fig. 1. Of particular interest are the measurements
of a pipe attached to the contraction exhibited the four downstream of a trip ring, reported in [3] and [6], which
corners of the chamber. When the chamber (whose filter mats differ substantially from those for natural flow development
acted as flow straighteners) was omitted, the data showed a made by the same researchers. Sharan [6] attributed this to
large scatter, because damping of the entry vortex was in- differences in turbulence structure. In the naturally
complete owing to screen instabilities. The disturbances were developing flow, he noted, in conjunction with Barbin and
particularly great when the screens also were removed so that Jones [10], that the turbulence energy in the boundary layer is
the intermittent entry vortex could pass without being dissipated in the axial direction rather than being convected
disturbed. All these macroscopic flow irregularities were radially outward. Contrary to this, the trip ring produced
avoided when the square chamber was replaced by an large eddies of high intensity close to the wall, which con-
axisymmetric one into which the flow entered axially through vected turbulence energy outward at a high rate. It appears to
a round filter plate. However, this gave rise to a new be the radial momentum transfer thus created which is
phenomenon, a fine transverse structure of the boundary responsible for the flow to develop at a much slower rate
layer (circumferential dynamic-head changes) in the con- initially than an undisturbed flow and for the centerline
traction which was magnified by the tripping device and velocity not to attain a maximum. The high development rate
persisted along the entire length of pipe (20 diameters). obtained by Liepe and Jahn [17] behind a screen ring (see the
Consequently, the boundary layer displayed a waviness at the data point in Fig. 1) may be due to particularly small eddies
circumference with departures of its thickness of 25 percent thus produced.
and of the wall shear stress of 10 percent from their mean The phenomenon can be also analyzed from Bradshaw's
values. The type of modulation changed when the screen [18] explanation for the peak in naturally developing flow.

Journal of Fluids Engineering JUNE 1981, Vol. 103/245

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 08/12/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


b-i

Tur
[%1 I
r ^
-<
From Poiel [14],
adjusted for Re 0 = M-10S:

bleed open
developed value far downstream. This is in line with Barbin
and Jones's observation of centerline intermittency starting at
x/D = 25 and boundary-layer merger at about x/D = 28.
/ Sale's distribution is completely different, having its
id < .
s* wy
'y

steepest gradient right at the pipe entry and having no
overshoot. This could be due to the crude pipe joints, but also
^ to the particular entry conditions: obviously the boundary
layer had already joined at the diffuser exit and therefore
n- turbulence was high in the entire cross section at this point2.
20 30 50 60 90 100 110 120
x/D Sprenger's [20] results, also obtained behind a contraction
Fig. 3 Development of cenierline turbulence intensity with non- preceded by a diffuser, indicate the same trend as Sale's.
dimensional distance from inlet Laufer's [21] experiments, for a long time believed to be
representative of fully developed flow, employed far too short
Weir, Priest and Sharan [3], ReD = 4.3.10 a pipe (x/D = 50). Patel [14], who was the first to mention
Sale [5], ReD = 4.0-10 this, did not obtain fully developed turbulence conditions
Miller [9], ReD = 5.5-105 even after 142 pipe diameters. Tuc increased by nearly 10
Barbin and Jones [10], ReD = 3.9105
Sprenger [20], ReD = 3.6-105 percent when a bleed upstream of the pipe was opened,
probably owing to sound waves transmitted from there. In
Sharan's [6] trip-ring experiments Tu at the center exceeded
the values for natural flow development within the first 30
When the wall boundary layers meet, the radial shear-stress diameters, but close to the wall for only 15 diameters. On the
distribution in either of them is very similar to that of a flat- other hand, at x/D = 65 Tu differed only slightly at the
plate boundary layer. Because of the type of curvature of this center, while close to the wall it was still lower than in
distribution, the shear-stress gradient over most of the central naturally developing flow even atxAD = 140.
portion exceeds that of fully developed pipe flow, in which the It is also worth mentioning that in the experiments without
shear stress changes linearly with the radius. As a con- a trip ring in [3] and [6], an increase of free-stream turbulence
sequence, the flow in the central region is accelerated. Thus intensity had no apparent effect on velocity-profile
the absence of a maximum centerline velocity in the trip-ring development at ReD = 4105 (see Fig. 1), in agreement with
tests suggests that the core-shear-stress gradient at boundary- tests reported in [14], but it did have an effect at ReD =
layer merger must have been smaller than in fully developed 1 105, a result confirmed by Lee and Goel [22]. This is due to
flow. According to Sharan's measurements this indeed is true, the transitional behavior of the boundary layer, which is also
while near the wall the shear-stress gradient was greater than the reason for the abrupt changes in slope at small values of
in fully developed flow. Although shear-stress and turbulence x/D displayed by some of the curves in Fig. 4. Some other
are related to each other, the former provides a deeper insight, distributions also show changes in slope at larger values of
because its slow adjustment in the central portion is x/D where, with the high Reynolds numbers used, the flow
responsible for the long distances required for velocity-profile must have been turbulent: Sprenger's [20] between x/d*=ll
development. In the wall region the adjustment rates of and 17, that of Jeng-Song-Wang and Tullis [19] between x/D
naturally developing flow are very much faster; the wall shear = 7 and 10, Sale's [5] at x/D~\2 and Deissler's [4] at
stress reaches its fully developed value after 8 to 15 pipe x/D 20 (Fig. 2). The reason remains unknown, but it is
diameters (results according to [6] and [10]), the pressure interesting to note that these are the cases where no overshoot
gradient after about 10 diameters [10]. of the centerline velocity occurs in the absence of trip rings, in
Apparently, roughness has a similar effect as a trip ring spite of well-rounded, large-area-ratio contractions being
[19], Fig. 2, with higher blockage factors at large distances fitted ahead of the pipe inlets.
(see the value of B = 0.20 at x/D = 96, from [13]). However,
this would not appear to explain Deissler's [4] measurements
made downstream of a bellmouth entry and Sale's [5] behind Laminar-turbulent Transition
a well designed 20:1 contraction. The curves obtained by these Some of the differences depicted in Figs. 1 and 2 are due to
two researchers are surprisingly similar. Sale incorporated a laminar-turbulent transition. This can be seen more clearly in
long 5 deg conical diffuser upstream of the contraction, and Fig. 4 with its enlarged x/Z?-scale where, in addition to curves
this could be the cause of the shape of his curve. On the other from Figs. 1 and 2, results of researchers who tested only
hand Sale reports that his pipe had crude joints between its short pipes have been included. Most distributions, when
sections, and this roughness could also account for the lack of extrapolated to x/D = 0, would have a negative value of B
"overshoot." Deissler's set-up employed a settling length (for example that of Robertson and Ross [23]). Since B must
upstream of the bellmouth, preceded by an orifice, a filter and remain positive, the slopes must change drastically at small
control valves. It is possible that the settling length was too values of x/D, indicating a significantly reduced development
short to fully eliminate the disturbances created upstream. It rate due to laminar flow in the initial portion of the pipes. In
is interesting to note that the distributions obtained when the fact, these curves have a pronounced kink, as is verified by the
bellmouth was replaced by a square-edged entry are very results of Sprenger [20] and of Muller [24]. The extent of
similar to the curves in Fig. 1 and therefore have been in- laminar or transitional flow portions depends primarily on
cluded here instead of in Fig. 2. the Reynolds number, but also on upstream conditions; in
Turbulence data for developing pipe flow would be of great particular, on the contraction ratio which determines the
help for a better understanding of the phenomena involved. degree of relaminarization of an initially turbulent boundary
Such data are scarce, however. The development of Tuc along layer. Contraction shape and length are of importance as well
a pipe downstream of a 16:1 contraction was measured in [3] as the presence or absence of screens which, in addition to
and compared with Sale's and Barbin and Jones's data. These affecting turbulence, may determine whether separation
and a few additional results are reproduced in Fig. 3. In three occurs in the contraction and thus affect the turbulence
of these cases Tuc remains fairly constant to about 15 to 25
diameters, then rises steeply, has a peak at x/D = 40 to 45 (at
approximately the same location where the centerline velocity 2
Note that Tuc = 1 percent was measured at the pipe entrance, which is an
has a maximum), and then falls very slowly towards its fully extremely large value for a 20:1 contraction.

246/Vol. 103, JUNE 1981 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 08/12/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


10 12 14 16 18 20 22 X /Q 26

Fig. 4 Development of the velocity profile within the first 26 pipe


diameters

Symbol Ref. 10" 5 X Tuc [*] Contr.


No. at entry ratio Remarks

O 20 3.5 0.45 25:1 diffuser upstream


23 12 ? 9:1 water tunnel
24 1.8 1.0 18:1
> 28 2.5 ? 4:1 conical contr.,faired
29 10 ? ? also: R e D = 20-10 5
e
All data points obtained from 6 (ax /D.
Further symbols as in Figs. 1 and 2

0,1* 5 to 8 pertains to natural transition without artificial means to


6 trip the boundary layer.
t/j.
As expected, the changes in slope of the curves in Fig. 4
D shift towards larger distances x/D when Re^ decreases (Fig.
,m
5). These distances are far larger than the ones at which the
.-*'" shape factor Hl2 starts falling towards its turbulent value; see
// ___
the arrows in Fig. 6. The beginning of this fall therefore in-
dicates much earlier transition than the slopes of the blockage
- .- curves. To clarify this observation, B and Hl2 were plotted in
I'M S- Figs. 7 and 8 against Re^., the Reynolds number defined with
distance x from pipe entry. The distributions in Fig. 7 are
similar to the well-known relationship for the flat-plate
o- friction factor3. In the log-log plot used, in both the laminar
14 and turbulent flow regimes B falls linearly as Re^. rises, the
x/D slope being smaller for turbulent flow. Respective laminar-
Fig. 5 Effect of Reynolds number on blockage-factor development, and turbulent-flow lines are joined by a curve representing the
from Sprenger [20]
transitional flow regime, which makes a pronounced kink
with the former and converges smoothly with the latter. The
Symbol A X o A V o locations of the kinks have been included in Fig. 5 (flagged
Re -10"
D
5
0.35 0.46 0.56 0.77 1.12 1.75 2.50 3.54 5.20 6.25 symbols) and are seen to coincide very well with the changes in
slope observed there. For all pipe lengths they occur at the
same value, Rej. = 6105 (Fig. 7), the lower critical Reynolds
Flagged symbols: onset of transition, from Fig. 7 number. The upper critical Reynolds number (where the
transitional curves become tangential to the turbulent-flow
structure at its outlet (see Klein, Ramjee and Venkataramani lines) has also one fixed value, Rex = 1.85106 4 , for all the
[25]). pipe lengths. In terms of pipe-Reynolds number, the begin-
A thorough experimental investigation of transitional ning of the turbulent regime thus ranges from Refl = 1.1 105
6
developing pipe flow was conducted by Sprenger [20]. It for*/Z>= 16.85 to ReD = 2.2.10 for x/D = 0.85.
appears that these results have not been sufficiently ap- Cross reference between Fig. 7 and Fig. 8 is facilitated by
preciated, probably because Sprenger presented them only in numbers which mark corresponding points. It can be seen
tables and not graphically. Therefore they have been plotted
in a comprehensive form and are depicted in Figs. 5-8. While 3
This is not surprising because this friction factor varies with the same power
various types of trips for initiating transition were used in the of Reynolds number as the displacement thickness.
investigations described so far, the discussion related to Figs. 4
This value was determined from a plot on a larger scale than Fig. 7.

Journal of Fluids Engineering JUNE 1981, Vol. 103/247

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 08/12/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


0,2 0,4 0,6 1 20 w
10- 5 -Re v
Fig. 7 Relationship between blockage factor and flow-distance
Reynolds number, from Sprenger [20]

Fig. 6 Effect of Reynolds number on shape-factor development, from Symbol + o <\ X V A 0 D o


Sprenger [20]
x/D 0 . 8 5 1.85 2.85 4 . 8 5 6.85 8.85 10.85 14.85 16.85
Symbols as in Fig. 5.
Arrows: location of minima of the curves B = f (Re x ) (Fig. 7)

x/D = 1.85, trip ring of height h/D = 0.005 a t x / D = 0.35


from the shape-factor distributions that neither the lower nor Encircled numbers and letters: for cross reference with Fig. 8
the upper critical Reynolds number can be deduced. One Arrows: termination of transitional and beginning of turbulent regime
could argue that determination of the latter from Fig. 7 was
inaccurate and that it must be assumed to be as low as 2,4
Re v = 1.0* 10 6 , where the shape-factor curves, after their steep ^ *\ ,^
fall, become nearly horizontal. However, this is very unlikely
nft v>
for the following reason: Sprenger tested a diffuser over a
range of Reynolds numbers downstream of various entrance
pipes. When plotted against blockage factor, all the non-
dimensional pressure recoveries for turbulent approach flow
2,0

i,a
CDC
r /
(B)
5& f\
/ ,

Lz

/ P
J8>

\
/p
formed a single curve, while recoveries for nonturbulent flow
departed from this curve. Now, for conditions corresponding 1,6 \ @> /
to points 17 and 21, which in Fig. 8 mark the end of the fall of
H{ 2, the pressure recoveries still indicate a nonturbulent
approach flow (see Fig. 10 of [7]), but turbulent flow for
1,4 "Z^
C D
<-*r.

points 18 and 22, which are close to the upper critical 1,2

Reynolds number, Re x = 1.85-106. This is even more ob-
vious in the tests where a trip ring was fitted 1.5 diameters 1,0
0,2 0,4 0,6 1 10 20 40
upstream. Here the blockage factor changes smoothly with Unite 100
Re^.. The lower critical Reynolds number is reduced to
Fig. 8 Relationship between shape factor and flow-distance Reynolds
Re^ = l10 5 . The upper critical Reynolds number is not number, from Sprenger [20]
clearly determined, but according to Fig. 10 in [7] it must be Symbols as in Fig. 7
Re A .>610 5 , because only then does diffuser performance Encircled numbers and letters: for cross reference with Fig. 7
correspond to turbulent flow (see points G and H). A much
lower limit for the termination of the transitional regime with 610 5 , the critical pipe length was x/D = 2.6 compared
would follow from the shape factor if the end of its steep fall with 13. These differences are due to the fact that Lee and
is regarded as a valid criterion (point D in Fig. 8). Goel did not derive the critical values from the blockage
factor, but from a plot of the friction factor against x/D5.
We must deduce that with the nonequilibrium boundary
One must conclude that not only does turbulent flow develop
layers produced by contractions upstream, the displacement
more quickly, but also that the beginning and end of tran-
thickness or blockage factor reflects the peculiarities of
sition occur much earlier at the wall region than at the central
transitional developing pipe flow far better than the shape
region. This has an analogy in the fully developed pipe flow
factor Hl2 or the momentum thickness. The two latter
tests of Patel and Head [26], where the upper critical
parameters are obviously more affected by differences in
Reynolds number was about four times higher when deter-
contraction geometry and by other upstream conditions. This
mined from blockage than when obtained from skin friction,
can be seen when comparing Sprenger's [20] and Muller's [24]
a result confirmed in [27]. Since, as was shown, Sprenger's
results for Re D = 1.8-105: While within the first five
diffuser data correlate very well with the critical quantities
diameters of flow development the blockage factors are nearly
derived from the blockage factor, the definition of critical
identical (Fig. 4), the shape factors are entirely different (Fig.
quantities by blockage is more reasonable than use of the
6). Consequently, for diffusers with entrance pipes, the
friction factor.
blockage factor is a more reasonable parameter than
momentum thickness for describing inlet conditions.
Lee and Goel [22] also investigated transitional developing Conclusions
pipe flow. They found a strong reduction of both the lower The conclusions from this review are as follows:
and the upper critical Reynolds number as the free-stream
turbulence intensity was increased. Both these quantities had 1 The development of both the velocity profile and tur-
lower values, and the pipe lengths required for the occurrence bulence in turbulent pipe flow depends largely on upstream
of transition were much shorter, than in Sprenger's tests. flow history. Undisturbed entry flow generates a maximum in
With Sprenger's turbulence level Tu c = 0.45 percent, their
lower critical Reynolds number was Re x = 1 105 compared 5
Velocity-profile development was not presented in [22].

248/Vol. 103, JUNE 1981 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 08/12/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


profile-peakiness at about 40 pipe-diameters, while certain Flow. A Guide to Losses in Pipe and Duct Systems," B.H.R.A.-Fluid
types of initial disturbances produce a smooth profile Engineering, Cranfield/Bedford, 1971.
development without such a maximum. 10 Barbin, A.R., and Jones, J.B., "Turbulent Flow in the Inlet Region of a
Smooth Pipe," ASME Journal of Basic Engineering, Vol. 85, No. 1, 1963, pp.
2 The distance required for full flow development may 29-34.
exceed 140 pipe diameters. Some of the classical 11 Cockrell, D.J., and Markland, E., "The Effect of Inlet Conditions on
measurements of fully developed pipe flow were in fact Incompressible Fluid Flow Through Conical Diffusers," Journal of the Royal
conducted at conditions when the flow was still developing. Aeronautical Society, Vol. 66, No. 1, 1962, pp. 51-52.
12 Bradshaw, P., "Review-Complex Turbulent Flows," ASME JOURNAL OF
3 The beginning and termination of laminar-turbulent FLUIDS ENGINEERING, Vol. 97, No. 2, 1975, pp. 146-154.
transition is determined most reasonably from a plot of the 13 Bradley, I.C., and Cockrell, D.J., "The Response of Diffusers to Flow
blockage factor against Reynolds number defined with flow Conditions at Their Inlet," Symposium on Internal Flows, University of
distance. For this, the momentum thickness or the shape Salford, England, 1971, Paper 5.
14 Patel, R.P., "A Note on Fully Developed Turbulent Flow Down a Cir-
factor are unsuitable parameters. cular Pipe," Aeronautical Journal, Vol. 78, No. 2/3, 1974, pp. 93-97.
4 The effects of inlet conditions on flow development are 15 Nikuradse, J.,"Gesetzmassigkeiten der turbulenten Stromung in glatten
not yet sufficiently understood. Rohren," Forschungsheft 356, VDI-VerlagGmbH, Berlin, 1932.
16 Pozzorini, R., "Das turbulente Stromungsfeld in einem langen
Kreiskegel-Diffusor," PhD dissertation ETH Zurich, 1976. Eduard Truninger
Note AG Zurich, 1976.
17 Liepe, F., and Jahn, K., "Untere Wirkungsgrade von Kegeldiffusoren,"
One of the reviewers pointed out three papers to the author Maschinenbautechnik, Vol. 11, No. 11, 1962, pp. 588-589.
which have been added to the list of references. 18 Bradshaw, P., An Introduction to Turbulence and its Measurement,
Walklate et al. [30] present blockage factors for Re D = Pergamon Press, New York, 1971.
2.25* 105 to x/D = 30 which nearly coincide with those ac- 19 Jeng-Song-Wang, and Tullis, J.P., "Turbulent Flow in the Entry Region
of a Rough Pipe," ASME JOURNAL OF FLUIDS ENGINEERING, Vol. 96, No. 1,
cording to Bradley [2], shown in Fig. 1. Turbulence intensities 1974, pp. 62-68.
from [30] agree well with the values according to [3] and [10] 20 Sprenger, H., "Experimentelle Untersuchungen an geraden und
in Fig. 3. gekriimmten Diffusoren," Mitteilung Nr. 27 aus dem Institut fur Aerodynamik
Reichert and Azad [31] present velocity profile develop- an der ETH Zurich, Verlag Leemann, Zurich, 1959. See also: "Experimental
investigations of straight and curved diffusers," British Ministry of Aviation
ment to x/D = 67 for two Reynolds numbers, Re D = Translation, TIL/T. 5134, 1962.
0.76'10 5 and 1.53-105. They used a 89:1 contraction cone 21 Laufer, J., "The Structure of Turbulence in Fully Developed Pipe Flow,"
and a 0.5-D long sandpaper trip at the pipe entry. Owing to NACA T.N. 2954, 1963.
this sandpaper, the initial boundary layer was extremely thick 22 Lee, Y., and Goel, K.C., "Free Stream Turbulence and Transition in a
Circular Duct," Proceedings of the Second International JSME Symposium
(B = 0.08 at x/D = 2). Apart from this, their blockage factor Fluid Machinery and Fluidics, Tokyo, Sept. 1972, pp. 11-20.
distributions are very similar to those of [3] and [11] shown in 23 Robertson, J.M., and Ross, D., "Effect of Entrance Conditions on
Fig. 1. They have marked peaks at x/D = 32 and minima at Diffuser Flow," Proceedings of the ASCE, Vol. 78, Separate No. 141, 1952,
x/D = 60. pp. 1-24.
24 Miiller, H.P., "Experimentelle Untersuchungen an Kegeldiffusoren
Laws et al. [32] present some evidence for an oscillatory einer mehrstutzigen Radialverdichterspirale," Brennstoff-Warme-Kraft, Vol.
development of pipe flow which has not been reported before. 28, No. 3, 1976, pp. 97-101. See also: PhD dissertation of same title, Ruhr-
Universitat Bochum, 1974.
25 Klein, A., Ramjee, V., and Venkataramani, K.S., "An Experimental
References Study of the Subsonic Flow in Axisymmetric Contractions," Zeitschrift fiir
Flugwissenschaften, Vol. 21, No. 9, 1973, pp. 312-320.
1 Schlichting, H., Boundary Layer Theory, 6th edition, McGraw-Hill,
26 Patel, V.C., and Head, M.R., "Some Observations on Skin Friction and
1968.
Velocity Profiles in Fully Developed Pipe and Channel Flows," Journal of
2 Cockrell, D.J., "Effect of Inlet and Outlet Conditions on Pipe and Duct
Fluid Mechanics, Vol. 38, Part 1, 1969, pp. 181-201.
Components,'' The Ninth Members Conference, Sept. 1967, Cranfield, SP 929,
The British Hydromechanics Research Association. 27 Pennell, W.T., Sparrow, E.M., and Eckert, E.R.G., "Turbulence In-
tensity and Time-Mean Velocity Distributions in Low Reynolds Number
3 Weir, J., Priest, A.J., and Sharan, V.K., "Research Note: The Effect of
Turbulent Pipe Flows," International Journal of Heat and Mass Transfer, Vol.
Inlet Disturbances on Turbulent Pipe Flow," Journal of Mechanical
15, 1972, pp. 1067-1074.
Engineering Science, Vol. 16, No. 3, 1974, pp. 211-213.
28 Winternitz, F.A.L., and Ramsay, W.J., "Effects of Inlet Boundary
4 Deissler, R.G., "Analytical and Experimental Investigation of Adiabatic
Layer on Pressure Recovery, Energy Conversion and Losses in Conical Dif-
Turbulent Flow in Smooth Tubes," NACA T.N.2138, 1950.
fusers," Journal of the Royal Aeronautical Society, Vol. 61, No. 2, 1957, pp.
5 Sale, D.E., "Entrance Effects in Incompressible Diffuser Flow," M.Sc. 116-124.
thesis, University of Manchester, 1967. 29 Little, B.H., and Wilbur, S.W., "Performance and Boundary-Layer
6 Sharan, V.K., "An Experimental Investigation of Naturally Developing Data From 12 and 23 Conical Diffusers at Area Ratio 2, 0 at Mach Numbers
Turbulent Flow and Flow with Fixed Transition in a Parallel Pipe," ASME- up to Choking and Reynolds Numbers up to 7, 5 X 1 0 6 , " NACA-Report 1201,
Paper 72 WA/FE-38, 1972. See Also: "The Effect of Inlet Disturbances on 1954.
Turbulent Boundary Layer Development in a Parallel Pipe," Journal of Ap- 30 Walklate, P., Heikal, M.R.F., and Hatton, A.P., "Measurement and
pliedMathematics and Physics fZAMP), Vol. 25, 1974, pp. 659-666. Prediction of Turbulence and Heat Transfer in the Entrance Region of a Pipe,''
7 Klein, A., "Review: Effects of Inlet Conditions on Conical-Diffuser Heat and Fluid Flow, Vol. 6, No. 2, 1976, pp. 89-95.
Performance," ASME JOURNAL OF FLUIDS ENGINEERING, published in this issue
31 Reichert, J.K., and Azad, R.S., "Nonasymptotic Behavior of Developing
pp.250-257.
8 Sovran, G., and Klomp, E., "Experimentally Determined Optimum Turbulent Pipe Flow," Canadian Journal of Physics, Vol. 54, No. 3, 1976, pp.
Geometries for Rectilinear Diffusers with Rectangular, Conical or Annular 268-278.
Cross Section," in Fluid Mechanics of Internal Flow (ed.: G. Sovran), Elsevier 32 Laws, E.M., Lim, E-H., and Livesey, J.L., "Turbulent Pipe Flows in
Publishing Company, Amsterdam, 1967. Development and Decay," Symposium on Turbulent Shear Flows, London,
9 Miller, D., "Performance of Straight Diffusers," Part II of "Internal June 1979.

Journal of Fluids Engineering JUNE 1981, Vol. 103/249

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 08/12/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

S-ar putea să vă placă și