GJ

m
-1

NATIONAL ADVISORY COMMITTEE

FOR AERONAUTICS

TECHNICAL NOTE 3066

EFFECT OF SURFACE DOWNSTREAM REGION

OF A 23 CONICAL DIFFUSER

By Jerome Persh and Bruce M. Bailey

,. Langley Aeronautical Laboratory

v Langley Field, Va.
i
.,

Washington
January 1954

!
II

----- ... ...- -_ .. . .. . . - . . ... . _. ___ . . . . . ... .. . .. --.-.. ...-= ~,. - ..-.
 TECH LIBRARYKAFB,NM

NATIONAL ADVISORY C@MIT%3E FOR AERONAUTICS

Iullllllllullllllllllln
llllbb27L

TECHNICALNOTE3066

EFFECT OF SURFACE ROUGENESS OVER THE DOWNSTREAM REGION

OF A 23 CONICAL DWFUSIZR

Ey Jerome Persh and Rruce M. Bailey

SUMMARY

An experimental investigation was conducted to determine the effects

of varying extents of surface roughness over the downstream region of a
23 conical diffuser having an inlet-boundary-lsyer thickness of the
order of 5 percent of the inlet diameter. The air flows used in this
investigation cover an inlet Mach nuuiberrange from about 0.10 to 0.40
corresponding ti Reynolds numbers of approximately 1 x 106 to 4 x 106
based on inlet diameter. The surface roughening was accomplished by
coating the surface of the diffuser with graded cork psrticles of a con-
trolled size. Incremental bands of roughness were removed fran the
upstream end (a l-inch-wide band being retahed near the inlet to sta-
bilize the flow) titer each series of pressure measurements was made so
that the variation of diffuser performance with percent of diffuser
length roughened could be determined.

The results of thepresent investigation and those of WARM L51K09

indicate that the flow in the roughened diffuser was steady and reproduci-
ble for all conditions. The values of total-pressure-loss coefficient
measured at both the tailpipe exit and diffuser exit for the almost fully
roughened diffuser were found to be 8 and 21 percent lower> respectively
than the value of total-pressure-loss coefficient for the smooth-surface
diffuser measured at the tailpipe exit.

INTRODUCTION

Reference 1 indicated that, although the static-pressure recovery

of a short, wide-angle diffuser was barely affected as a result of either
roughening almost the entire surface of the diffuser or installing a
roughness strip near the diffuser inlet, the flow pattern was consider-
ably improved over that found in the same diffuser wfthout any roughness.
This encouraging result suggested that, since a steady symmetrical flow
pattern could be achieved by using surface roughness, possible g- in

-.. . .. . .. . .. .c _ -z ..
2 NACATN 3066

diffuser performance might be realized by jU&M,OUS placement of the areas

of surface roughuess.

The skin friction in the upstresm region of a diffuser probably con-

tributes more to the overall total-pressure losses than does the skin
friction in the downstream region because the boundary layer approaches
a separated condition as it flows toward the diffuser exit, with an accom-
PSW@-W3 decrease in skin-friction coefficient. Speculation that tie per-
formance of the almost fully roughened diffuser (ref. 1) might be improved
by decreasing the skin friction in the upstream region of the diffuser,
consequently, led to the expediency of increasing the extent of smooth
surface in this region. Byrets..ng the roughness strip near the dif-
fuser inlet, this procedure couldbe acccqlished without sacrificing
the flow stability.

The present investigation was therefore undertaken as a continua-

tion of the investigation of reference 1 to determine whether gains in
diffuser performance could be attained through the judicious use of con-
tinuous surface roughness. The results of the present investigation, in
which the variation of diffuser performance with extent of roughness
increasing from the diffuser exit in an upstream direction is detemnined,
are directly comparable with the results of reference 1, inwhichthe
variation of diffuser performance was studied with extent of roughness
increasing from a point near the inlet in a downstream direction, because
essentially the ssme apparatus was used for both investigations and the
thickness and shape factors of the inlet bouudary lsyer were the same
for both experiments.

The data presented herein cover an inlet Mach number range from
about 0.10 to o.ko corresponding to Reynolds numbers of approximately
1 X 106 to 4 X 106 b~ed on wet ~~e~r. me e~ent of romss
over the downstream region of the diffuser was varied so that the varia-
tion of diffuser performance with percent of diffuser length roughened
could be computed. Boundary-layer velocity proffles are presented for
all rouglness configurations as obtained frum measurements at the dif-
fUSer inlet station, several lo~~dinal points in the diffuser, the
diffuser exit, and the tailpipe tit.

SYMBOLS

t3 acceleration due to gravity, ft/sec2

h total pressure, lb/sq ft

E weighted mean value of total pressure, lb/sq ft

WA TN 3066 3

m weighted-total-pressure loss,from pressure surveys, lb/sq ft

L diffuser length, in.

M Mach nwiber

P static pressure, lb/sq ft

AP wall static-pressure rise, lb/sq ft

Pideal
static-pressure rise for frictionless, incompressible, one-
dimensional flow with ssme entering mass flow and geomet-
ric sxea, lb/sq ft

bsmmetric pressure, in. Hg

o

To stagnation temperature, %

impact pressure, ~ - p, lb/sq ft

%

r radisl distsnce from center line, in.

R radius, in.

% Reynolds number based on inlet diameter, *=

P

u local.velocity at any point, ft/sec

u local velocity at edge of boundary layer, ft/sec

I
u/u velocity ratio,
L:P
J

Ws standardized weight flow,

&W&= i, b/see

x distance along longitudinal axis measured from inlet,

station 1, in.

Y perpendicular distance fran diffuserwdl, in.

IJ viscosity, lb-sec/ft2

.-_ -.
4 NACATN 3066

mass density, lb-sec2/ft4

boundary-layer thickness, equal to y for ~ = 0.95, in.

boundary-layer displacement thickness fQr incompressible

e boundary-layer momentum thickness for incompressible

lm iw-$(:)
H boundary-layer shape parameter for incompressible
flow, */e

Diffuser performance psr-ters:

m total-pressure-loss coefficient
q

4 diffuser effactiveness
4ide~

Subscript%:

0 reference conditions

1 diffuser-inlet conditions

6 diffuser-exit conditions

7 tailpipe-exit conditions

max maximum value

x longitudinal distance along length of diffuser

APPARATUS AND TESTS

b

General arrangement.- A schematic drawing of the apparatus used for

this investigation is shown in figure 1. The test duct system consists
of a 23 conical d3.ffuserwith a 2:1 ratio of exit to inlet area joined

.
 NACATN 3066 5

to a 21-inch-cWmeter cylindrical approach tube approximately ~ inlet

diameters in length. The junction between the approach tube and dif-
fuser was formed as a circular arc of ~ - inch radius, tangent to both
the inlet cylinder and diffuser cone. ~discharge tailpipe approxi-
mately ~ inlet diameters in length was attached to the diffuser etit.
For all configurations, a l-inch-wide roughness strip was permanently
installed near the diffuser inlet to stabilize the flow. This strip is
shown in figures 2 and 3.

Roughness particle size.- The cork particles of which the surface

roughness was composed were the same size as those used for the inves-
tigation of reference 1. These particles will pass through a standard
screen with 8 meshes to the inch but wiXL be retained on a standard
screen with 14 meshes to the inch. The average height of the particles
used is approximately 0.10 inch. The data of reference 2 indicate that
the aversge skin-friction coefficient for roughness made up of these
particles is 0.0035 for a pipe. This value is about three times the
valw of the average skin-friction coefficient for the Smooth+urface
diff@er (ref. 3) over the range of Reynolds numbers investigated.

Description of configurations.- The cork particles were uniformly

cemented about the interior surface of the diffuser in various extents
of surface roughness. The leading edges of the roughness of the con-
figurations are shown in figure >(a) and the configurations are desig-
nated V, VI, VII, and VIII to conform with the designation system used
in reference 1. The first series of pressure messurements was made for
configuration V which had about 86 percent of the diffuser length ro@-
ened. Succeeding configurations VI, VII, and VIII had appro-tely 70,
54, and >2 percent of the diffuser length roughened, respectively. The
l-inch roughness strip near the inlet was included for W computations
of percentages of diffuser length roughened. ~cremental bands of cork
were removed frm the upstream edge after each series of pressure meas-
urements was made so that the variation of the diffuser perfoma.nce
with percent of diffuser length roughened could be detemined. It should
be noted (fig. 3) that the extent of the smooth surface between the
trailing edge of the inlet roughness strip and the leading edge of the
main roughness region beccmes larger with the removal of each incremental
band of roughness. For each series of pressure measurements, t~ lea-
edge of the roughness was buffed and faired smoothly into the diffuser
wall to a point about 2 inches downstream of the leadlng edge.

Instrumentation and inlet calibration.- A series of static-pressuxe

orifices were installed along O* generatrix of the diffuser @ tail-
pipe to measure longitudinal static-pressure distributions. As is pointed
out in reference 1, the static-pressure me=uremerits obtained from orifices

.- . . ....-. ._ .
.

6 NACATN 3066

located in the roughened area are believed to be accurate. At sta-

tions 1, 6, and 7 (fig. 1), wall static-pressure measurements were made
at six eq@lly distributed positions around the circtierence. Pitot-
static-pressure surveys also were made at longitudinal stations 2, 3,
k, and 7 between the diffuser inlet and exit for purposes of studyztng
the boundsry-lsyer development. The locations of these stations. are
indicated in figure 1.

The flow conditions at the cliff

user imlet, station 1, were care-
fully explored and calibrated by making pitot-static-pressure surveys
at three equidistant points around the circumference of the inlet.
Typical velocity profiles measured at the diffuser inlet me shown in
figure 4 for several values of inlet pressure ratio p~pO. The results
of the diffuser-inlet calibration are shown in figure 5 in which the
inlet Mach number, the Reynolds number based on inlet diameter, the
weight flow adjusted for standard stagnation conditions of 29.92 inches
of mercury and 520 R, and a curve of pl/~ are all plotted as func-
tions of the inlet pressure ratio.

For all cotiigurations, three pitot-static-pressure surveys were

made simultaneously at three equally distributed positions in the trans-
verse plane of station 6, for the purpose of checking flow symnetry.

Weight-flow check.- !I&weight flows were calculated for each con-

figuration at both the diffuser efit and tailpipe exit. The results of
these calculations are shown in figure 6 in which the standardized
weight flow is plotted as a function of the inlet pressure ratio for all
configurations at stations 1, 6, snd 7. For all configurations the
weight flows calculated frmn pitot-static-pressure measurements at the
tiffuser @t, given in figure 6(a), are s~ghtly brger than the weight
flows calculated from such measurements at the Wet, especially at the
higher velocities. Discrepancies between inlet and exit weight flows
were noted in references 1 snd k, and reference 7 provides a method for
estimating the effect on the performance results of pressure measurements
which lead to such weight-flow discrepancies. An estimation made with
the use of the procedure of reference 5 indicated that for configura-
tions VI and VII, for which the weight-flow discrepancies are greatest
at the diffuser tit, the calculated values of diffuser total-pressure-
10SS coefficient, discussed in the section entitled Results and Dis-
cussion, are probably of the order of 5 percent lower than the true
mean values.

The measured weight-flow vslues for configurations V and VIII at

the diffuser ebt, station 6, and for all the configurations at the
tailpipe exit, station 7, indicate that sny attempted correction si.m-
ilsr to that of reference 5 for these cases would amount to less than
the data scatter. NO total--pressure-loss~coefficient corrections were
made, therefore, for any of the data presented.
NACA TN 3066

METHODS AND ANALYSIS

Because a reference -totsJ.-pressure

tube installed in a small plenum
chamber of the present apparatus would produce disturbances in the inlet
flow, the static pressure p. was used as the reference pressure. In
reference 1, the plenum chsmber was larger; therefore, a total-pressure
tube was installed and the inle$ total pressure ~ was used as the
reference pressure. Thus, ti comparisons between .thedata for the
present investigation and those reported in reference 1 were made by
using the calibration curve of figure 5(d) in which pl~ is plotted

Calculation of pressure differences.- The volume-weightedmean loss

in total pressure from the reference station O to the station under con-
sideration was camputed in the following manner:

o
+= JR - X)rm
(PO
-H
(1)

J
Y(
o
Urdr

The mean loss in total pressure was camputed for both the diffuser and
diffuser plus tailpipe by usi~ the following relations:

For the tiffuser:

.
%,6 (Po - q - (Po - @ (2)

For the dLffuser plus tailpipe:

&,7 (Po - ~) - (Po - %) (3)

The rise in static pressure was cmputed as the difference between

the arithmetic mean of t-h six wa12 static-pressure measurements at sta-
tion 1 and the arithmetic mesn of the wsll static-pressure measurements
at station 6 or 7. The theoretical gain in static pressure was cmputed
by assuming frictionless, incompressible, one-dimensional flow with the
same entering mass flow and gecmetric area.
f

.. ...
 ..

8 NACA TN 3066

Diffuser performance parameters.- W order to provide a basis for

cmparing the results of the present investigation with those of ref-
erence 1, the ssme performance parameters as presented therein are used
in the present analysis. The coefficients are given as follows:

(1) The total-pressure-loss coefficient, defined as the loss in

mesn total pressure divided by the inlet @act pressure, ~/9.cl

(2) The diffuser effectiveness, defined as the actual gain in static

pressure dividedby the gain in static pressure possible with friction-
less flow, 4/4~&s&

RESULTS ANDDIE4HJSSION

~ the.present investigation the ssme inlet-boundary-layer thick-

ness existed as in that of reference 1. This inlet-boundary-lsyer
thickess was of the order of 5 percent of the inlet dismeter and cor-
responds to the thicker inlet-boundary-layer condition of reference 3.
AU comparisons between the data presented herein and those of refer-
ence 1 or 3 are made for this inlet-boundary-layer condition.

Flow in Roughened Diffuser

In contrast to the flow in the smooth-surface diffuser (ref. 3)

which periodically shifted position and lacked reproducibility, the
flow in the roughened diffuser was steady and reproducible. Although
flow separation was found at the Uffuser exit for some configurations,
this condition was not characterizedby violent oscillations of the
fluid in the manometer tubes. Furthermore, it was found that the data
were rea&l.y repeatable for the cases in which lmundary-layer separa-
tion appesred.

As pointed out in reference 3, difficulty in making pressure sur-

veys at the diffuser exit made it impossible in that investigation to
present vslues of the total-pressure-loss coefficient at that point.
In both the present investigation and that of reference 1, however, the
steady flow at station 6 made it possible to make detailed pressure sur-
veys at that petit and values of, &~l are presented at station 6 for
all.configurations. The diffuser effec~iveness is the only performance
parameter for which an exact cmp.risen canbe made at the diffuser exit
(station 6) between the smooth-surface-diffuserresults of reference 3
and the present study.

.
P
NA.CATN 3066 9

Boundary-Layer and Diffuser Performance Results

For all configurations, the velocity profiles cmputed from pitot-

static-pressure surveys made at seven stations along the waU of the
diffuser and tallpi~e on a single generatrix are shown in figures 7
to 10 for three approximately constant values of pl/po. Jhcluded with
each of the profiles shown in figures.7 to 10 are the values of the
boundary-layer displacement and mmentum thicknesses and the boundary.
layer shape parameter. Velocity profiles at three eqtidlstant points
on the circumference in the plane of station 6 are shown in figure u
for an approximately constant inlet pressure ratio of 0.94. The
boundsry-layer parameters W, (1,and H, calculated fran test tits,
are plotted against the inlet pressure ratio in figures 1.2to 16, for
each of t& roughness configurations. VaI-uesof W, 0, and H
selected from the faired curves of figures 1.2to 16 at an inlet pres-
sure ratio of 0.95 are plotted in figure 17 to show the development of
the boundary layer along the length of the diffuser for each of the
roughness configurations.

Figure 18 shows a cmp~son between the static-pressure distribu-

tions for each of the roughness configurations at an approximately con-
stant inlet pressure ratio of 0.95. The variation of the diffuser
effectiveness and total-pressure-loss coefficient with inlet pressure
ratio are shown in figures 19 and 20, respectively, for dl fo~ of the
configurations investigated. A cmparison between the diffuser-exit
(station 6) velocity profiles for w configurations, including con-
figuration I of reference 1, is shown in figure 21. Figure 22 presents
a r&u and ccwqmrison between the performance results for the present
investigation and those of reference 1.

Diffuser boundsry-layer velocity profiles.- The results for the

boundary-layer velocity prpfile along the diffuser and at the tailpipe
etit are shown in figures 7 to 10 and indicate that, for each of the
rouglness configurations, flow separation occurred or appeared inminent
at the downstream measuring stations 5 and 6. The data shown in fig-
ure 7 indicate that configuration VIII (32 percent of diffuser length
roughened) produced separated flow frcxnabout x equsls 14.5 inches
to x ecpals 24.6 inches (stations 4, 5, and 6) at all.velocities. ~
profiles shown for configurations VIIj VI, and V (~, 70, and 86 percent
of diffuser length roughened, respectively) indicate different degrees
of imminent separation at stations 5 and 6. (See figs. 8 to 10.) These
results suggest the probabili~ of asymmetrical flow conditions in the
dmtresm regions of the diffuser for all configurations and caution
should therefore be exercised in interpreting the results in terms of
the standard boundary-layer parame+xrs.

An indication of the asymetry of the diffuser flow maybe obtained

from the measurements made along three equally spaced radii at the

..-. ._ .Z .- _______ . .- .-
10 NACA TN 3066

diffuser exit, station 6. These measurements are presented in figure 11

in terms of boundary-layer velocity distributions for each of the rough-
ness configurations investigated. For configuration VIXI (32 percent of
diffuser length roughened), large variations in profiles were obtained
with separated flow being indicated in one position, as noted in fig-
ure 7, and attached flow being indicated in the other two positions. In
genersl, as the extent of roughness was increased, the degree of asym-
metry tended to decrease, with configuration V (86 percent of diffuser
length roughened) exhibiting quite symmetrical.flow. The flow asymme-
tries obtained preclude detailed interpretation of the boundary-layer
measurements along any single diffuser generatrix according to boundRry-
layer theory for symmetrical flow. However, the general trends indi-
cated by the curves shown in figure 17 are considered realistic.

Bolmdary-@y er displacement thickness.- The curves shown in fig-

ure 17(a) indicate that the variation of the boundary-layer displacement
thick&~ along the length of the diffuser is approdmatel.y the same for
all configurations within about JO percent. It can be seen that the
values of 5* for configuration VIII (32 percent of diffuser length
roughened) at stations 4, 5, ad 6 (x = 14.5, 19.8, and 24.6 inches)
are smewhat higher than the values of 5* for the other co~iguratio~.
This result is probably due to the separated flow indicated in figure 7
at these stations.

Since the value of &$ at any point along the wall of the diffuser
determines the effective area of the duct at that point, it would be
expected that the tial static-pressure distribution would be approxi-
mately the same for all configurations. lhisconclusion is supported by
the results shown in figure 18, in which the ratio of static pressure at
points along the length of the diffuser to the static pressure at the dif-
fuser inlet is plotted against tistance along me longitudinal axis for
approximately the same inlet pressure ratio for all configurations. The
deviation of the data from a single curve results in a difference of
about 10 percent in the pressure recovery for the most divergent case.

Since the differences in static pressure due to changes in roughness

length are small, and since all configurations produced approximately the
s= 5* ~ation, it c~be concluded that changes inrouglmess length
would not produce significant changes in the overall static-pressure
recovery. TMS conclusion is substantiated by the curves of figure 22(a)
which show the variation of &Lffuser effectiveness with percent of dif-
fuser length roughened. Examination of these curves indicates that,
although the diffuser effectiveness diminishes slightly as the extent of
roughness is increased, the differences are very small. ,/

Boundary-layer mcrmentumthickness.- The variation of the boundary-

layer momentum thickness along the length of the di$fuser is shown in
figure 17(b) for all configurations. For configurations V, VI, and VII
WA TN 3066 11

(86, 70, and54 percent of diffuser length roughened, respectively), Ahe

momentum-thickness variation was very similar and differences between
these curves are not considered large enough to be regarded as signifi-
cant. The values of e for configuration VIII (32 percent of diffuser
length roughened) are considerab~ lower than those for the other rough-
ness confi~tionsj however, these values of e cannot be regarded as
representative values occurring at other circumferential locations in
the diffuser because of the flow asymnetry resulting frcnnthe separation
in the downstream regions.

Boundary-layer shape parsmeter.- ~ curves of boundary-layer shape

parsmeter shown in figure 17(c) indicate that for configuration Vthe
rate of growth of H is essentially constant over the diffuser length.
For configurations VI, VII, and VIII, the slopes of the H curves have ~
been caused to reverse in sense near the diffuser exit.

Diffuser-exit velocity profiles.- Although the values of H at the

diffuser exit vary from about 3.0 to 3.5 for the different configurations
(fig. 17(c)), it is not apparent whether these values represent large
differences in the velocity-profile shapes, unless the physical shapes
of the velocity profiles are compared. h order to determine whether
significant differences exist between the diffuser-efit velocity pro-
files for SU configurations, figure ~ ccmpares station 6 velocity
profiles at a constant inlet pressure ratio of 0.95. Significant dif-
ferences between velocity profiles for all configurations are detectable.

A cmp=ison between the data shown in figures 7 to 10 aud the data

of figure 21 in~cates some differences between the two sets of data.
These differences occur because the profiles of figure 21 represent an
average of the three radial surveys, shown in figure I-1,whereas the data
shown in figures 7 to 10 correspond to radial surveys at one circumfer-
ential location.

In deciding which of the configurations produced the best overall

perfomnance, dd. aspects being considered, one would have to choose on
the basis of the values of the total-pressure-loss coefficient and the
exit-veloci~ distribution since the differences in static-pressure
recovery were not significant. The configurations with 32 and 54 percent
of the diffuser length roughened produced the highest total-pressure-
10ss coefficients as we~ as flow asymmetries at the diffuser tit; thus,
these configurations me undesirable. Figure 22(b) indicates that the
total-pressure-loss coefficient for the configuration with 86 percent of
the diffuser length roughened is about 10 percent lower than that for
the configuration with 70 percent of the diffuser length roughened and
figure H(d) indicates that it had the best flow symetry characteristics.
Configuration V, therefore, appears to be the best from an overall per-
formance sts.ndpointin the present investigation.
12 NACA TN 3066

Comparison Eetween Present Results and Those of Reference 1

The results of the present investigation can be shown to be in

accord with those of reference 1. To illustrate this fact, figure 22
shows the variation of 4/4i~~ and A$/~l with percent of dif-
fuser length roughened for both the diffuser and the tailpipe etits,
stations 6 and 7, at a constant inlet pressure ratio of 0.95. A com-
parisonbe%ween the results of the present investigation and the results
of reference 1 is also shown in this figure. Figure 22(a) indicates
that the static-pressure-recoveryresults for the present instigation
and for the investigation of reference 1 are, in general.,very much the
same. Consequently, the displacement thicknesses for the diffuser-exl.t
velocity profiles are very similar in magnitude for both investigations.
This fact canbe seenby comparing values of & forthe diffuser-exit
velocity profiles given in reference 1 and the rsmge of values of &
shown in figure 17(a) at station 6 (x = 24.6 inches). Figure 22(b)
shows that, although the msdmum values of the total-pressure-loss coef-
ficient occur at approximately the same condition (45 percent of the
diffuser length roughened), the maximum values of m% at both meas-
/ 1
uring stations are somewhat less for the present investigation than those
found in the results of reference 1. This observation is readily explain-
able when it is considered that, for the investigation of reference 1,
the roughness extended from a~oint near the inlet in a downstream tirec-
tion, whereas the present results are for the case of roughness extending
from the diffuser exit in an upstream tiection. The exit-velocity pro-
file data of reference 1 inticate further that the 97-percent-roughness
configuration produced the best velocity distribution (see fig. 21).
The g7-yercent-rou@ness case also produced the smallest total-pressure-
10ss coefficients; values for the tailpipe exit and diffuser exit were
lower than that for the tailpipe exit of the completely smooth surface
diffuser by appr-tely 8 and 23. percent, respectiwly.

CONCLUSIONS

mom the investigation of NACA RM L51K09 and the present investi-

gation regarding the effect of surface roughness on the performance of
a 23 conical diffuser @th a 2:1 ratio of exit to inlet area and a
constant-area tailpipe ~ inlet diameters in length, and with sm inlet-
boundary-lsyer thickness of approximately 5 percent of the inl.et-dism-
eter, the followhg conclusions are drawn:
Y
WA~ 3066 13

1. The results of the present investigation and those of NACA

RM L51K09 indicate that the flow in the roughened diffuser (including
that for the inlet roughness strip only), even though asymmetrical for
acme configurations, was steady and reproducible for all conditions,
includ@ those with separated flow.

2. The 97-percent-roughness case produced the smallest total-

pressure-loss coefficients; values for the tailpipe etit and diffuser
exit were lower than that for the tailpipe exit of the completely smooth
surface diffuser by 8 and 21 percent, respectively. The ~e~n~
results indicated that maximum values of total-pressure-loss coefficient
at both the diffuser and tailpipe @ts were obtained for about 45 per-
cent of the diffwer len@h roughened, without regard to whether the
roughness was placed upstream or downstream, with the inlet roughness
strip in,place.

3. The results of the present investigation and those of NACA

RM L51K09 indicated a trend toward reduction of asymetry with increasing
extent of surface roughness.

4. The static-pressure recoveryat the diffuser @t diminished

slightly as the extent of the roughness was increased from the smooth
condition to the condition in which 97 percent of the diffuser length
was roughened, whereas the diminution of the static-pressure recovery
at the tailpipe exit was barely detectable over the range of roughness
configurations investigated.

Langley Aeronautical laboratory,

National Advisory Ccmmrltteefor Aeronautics,
~ey~eld, Vs., October 30, 1953.

- .. . - .- -.- - ~ ... .._.. - - ---

14 NACATN 3066

1. Persh, Jerome: The Effect of Surface Roughness on the Performance of

a 230 Conical Diffuser at Subsonic Mach Numbers. NACARM L51K09,
1952.

2. Nikuradse, J.: Laws of Flow in Rou@ Pipes. NACA TM u292, 1950.

3. Persh, Jerome: The Effect of the Inlet Mach Number and Inlet-
Boundary-Layer Thickness on the Performance of a 23 Conicsl-
Diffuser - Tail-pipe Canbination. NACARM LgaO, 1950.

4. Persh, Jerome, and Bailey, Bruce M.: Effect of Various Arrangements

of Triangular Ledges on the Performance of a 230 Conical Diffuser
at Subsonic Mach Numbers. NACA TN 3M!3, 1954.

5. Persh, Jerome, and Bailey, Bruce M.: A Method for Estimating the
Effect of Turbulent Velocity Fluctuations in the Boundary Layer
on Diffuser Total-Pressure-LossMeasurements. NACA TN 3124, 1954.
 Diffuser ~ Tailpipe

I \

Air flOW

To blwer
I I
I
Reference 6tatic-
L prassure orifice, Inlet , I&it, - Tailpips tit,
stition O station 1 atatlon 6 Etation 7

Ebundery-leyer survey stationa Four equally

stition 123456
/_ armed Btetic
0.030Malll,
3.?3
2.3 1-+=#m:is
- .@Odiem,
~ 15

%
Pitot Opsrli i6
0.023x O.cp 2
1/4 diem.
Cross section of diffwer PreBaum probe

Figure 1.- General. arrangement of test apparatus emd instrumentation.

All dimensions are in Inches.
16 WA ~ 3066
\

,.

L-71847.1
Figure 2.- Three-quarter view of diffuser, looldng upstream, showing
inlet roughness strip in place.

. .
 NACA TN 3066 17

E=O%L1
of flow

(a) Cross sectionof diffuser wall showing various surface-roughness

dimensions relative to inlet.

diffuser
Section of inlet duct IT

. -- .---. .- ---.:;
. . . , -.. -:-*.
....
.
....
. .......
. .-.::-::;
.
. .-::::?
. . . . .-:;.;
.
. .--
.-.....
J
Inlet roughness strip .
. .
,
.
.-..;
,.
..-..*-- -. . . . . .-,...--..,
flow
Roughnessparticles \w_$: :-----

(b) Isometric view of ~cross section of diffuser with extent of roughness

approximately representative of configuration V (85 percent of diffuser
length roughened).

Figure 3.- Details of diffuser surface roughness. All dimensions

are in inches.

-.. ._________ . ....-. . . . . ..

 18 NACA TN 3066

PI
~*,fi. g,~. H
&
0.9880 0.188 0.135 1.3go

.9792 .175 .133 1.320

.9600 .163 .128 1.275

.9100 .144 .llg 1.200

n
u I I w
I o
Olv
M PJ PI PI I
PI = 0.9600 q
= 0.9792 = 0.9100
= 0.9880 g Po
Lb %
c) \) ()
/.4-
() (: ) (:
)
u
40-
y, in
.8

.6 -
.4

I
Lo
J I I 1 I 1
o.z4.G .8@
i I I I
0.Z4.G.6L0 .
1 I 1 I 1 1
0,.2 .4 .6 .8 /.0
.

Figure 4.- Typical inlet velocity profiles at several values of inlet

pressure ratio.
 .
.*
..
X
%

.92 .90
+o
(a) Mach number.
~;. In&
x/u-

z
Re

< Dounsfxeamview of inlet showing

radial measuring stations
Lo .98 ,94 .94 .92 .90
P1/Po
.-
(b) Reynolds number.

Variation of inlet conditions with inlet pressure ratio.

_____ -._.,___ .. .. .
20

80

60

40
w lb
E

20

0
[0 .98 .96 . .94 .92 .90
P1/Pfj
(c) Weight flow.

.90

9Z

.94 -

.96

-1

/.0
10 .98 .96 , .94 .9Z .50
PIIPO-

(d) Ratio of inlet static pressure to upstream total pressure.

Figure 5.- Concluded.

 NACATN 3066 ZIl

80
Met calibration(station1)

g
60

El

Configuration
Zo Ov
VI
; ;I1
A

Lo .98 .96 .94 .92 .90

P1/Po

(a)rfiom diffuser-exit measurements (station 6).

80

60
El

40
wBY lb
=

20

o
/.0 .98 .96 .94 .92 .90
P1/Po

(b) IZromtailpipe-exit measurements (station 7).

Figure 6.- Weight-flow vsxiation with pl/pO from measurements at diffuser

and tailpipe exits.

..-..
--,--- . .
 /6 I I 1 I I 1 I
I I I
Sta. 6 I I

Sta. 4
Ek&2 ~ --y
/2 _ .
g Sta. 1 5* - 1.40

H = 4.34
(3
ill. H = 1.3

4 v

I n I
o
0 .2 4 .i,+lo g OJ?.4.6,8L0
I t 1.
u
0 .2 .4 .6 .8 10 O .2 .4 .6 ,8 !0
~~
o .Z .4,6, L3D0.Z .4.6 .8LO
O .2 ,4 .6 ,8 LO

Figure 7.- Rmm3my-1.ayer velocity profiles along diffw2er md at tailpipe

exit for 32 ~rcent of diffumm lengbh roughened (configuration VIII).
I I I 1 I I I 1 ,
I ma. 6 I
I
Sta. 5
.
S&9. 4
.
ti. 3
*.2 -
. . -
%ls~l

8
y, in.

,,
... .
0 I

Figure 7.- Continued.

 N
.!=

I I I I I I ! 1 1 I W. 6
I 1 Sta. 7
.Sta.
5 --
.
ah. 4
. Ha. 3 6- -
/2

8
e n O.l!a
H = 1.22
y, in.

1
0
I I u J I I
0. .2 .4 ,6 .8 /0 I
I 1
, , ,q.{. ~. L8/o
0.Z4. (6.8IDO,2.4,6L51O

O .2 .4 .6 .8 10

Figure 7.- Concluded.

*
 /6

-+1
r
I I !

Sta. 3
I I
.Sta. 4 .
I

St?..5
. 3

Y,
~F~
4
I
ii?
Pe
n- 1
0.Z4&.8~0 0 .~ .4 .6 ,8 [0
I I L + I I I
o .2 .4 .6 .8 [0 O ,2 .4 .6 .8 LO
~~
O. Z.4.6.6100I I .Z.4.6.8[O
I I I
I

0.24.6.0/0

(a) ~ M 0.96.

l?igure8.- E!Omdary-layer velocity profiles along diHumr and at tailpipe

exit for ~ percent of diffuser length roughened (configuration VII).
 /6 - I I I I I I I I 1 I
Sta. 5
W.
. 6- I
.
9ta. 7

/2

I
c St9. 1
-T!
I
1

yo.33
.
w.
-.
2
J
~1
r
Sta.4 L
-T

7: b+ _ 0.714 . 6* ~ 1.32 : 8* - 1.94

. $1

6* - 2.57 B* - l.m
7

! I 1/1) ii!
- 0 .214 ,9 .0.236 . e m 0.450 e n 0.582 , e . (3J394
e -1.26.
,H-2.!JO .H -2.93 H - 3.34 .E -2.87 H=l.38.

9 = 0.J24
Y, in. I H = 1.24 l.
L

6 I

Figure 8.- Continued.

9
I
lb [ I I I 1 1 I I I I M: 6 St-a.7 is
. ml. 3
h. 4

I -. ~ ~
.- -. Y w

Y,

I
0 .1 .4 .6 ,$ do
I 1 O 2 .4 .6 .8 /0
1 + 1 I I
0.24.6.6/0
1 1 { I I I o .2 Sf .6 .8 /0
0.Z.4,6.810 O. Z4.6.6[O

02.4 .6.8[0

Figure 8.- Concluded.

 /6

- . , ,,, ,,
1- I T Y 1 Y, r I II

y, in.

4 A A A A A d )4

0 I I
O .2 .4 .6 .8 1.0 I 8 0 .2 .4 ,6 .8 10
J + ) 8 1
O.z A.6.8L0 o .2 .4 .6 .8 /0
~~
0. Z.4.6.8[00I .Z4.6.8L0
I ! I 1 )
0 .2.4.6.8/.0

(a) ~ = 0.96.

Figure 9.- M_lnaary-layer velocity proffl,es along ditfuser .wd at tailpipe

exit for 70 percent of diffuser length roughened (configuration VI).
G

*
 i

I 8ta. 7
.

,
I
I I
Sta. 1
%8* = 1.56 .

d
y,ill.

0 I I %%4 I I I I
e
;8,0.f..f.6.i340
I -u

I
t
I Figure 9.- Coutj.nued.
 %

/6 I 1 I I I I
Sh. y
I WA,.
.. 6 .
1
Ma.
. T
1 9ta. 4
.

- B*. 3
Asta, 2. A f
k?
g,sta.
. 1 J J .
6* - 0,364 .
8 e = 0.194
,.,.f,, = 0.910A : -
8
y, b.

0,24.6.8/0,.,,, I i o ,Z ,4 ,6 .8 10
0..Z.4,6.8 iOO.Z.4.6.8LO
O .2 .4 .6 .8 [0

(c) $ = 0.915.

Flguce 9.- Concluded.

 /6 I I I I 1 I 1 1 I sta. 6 I E/
Sta. 9
.
SW. 4
ma. 3
.
Eta. 2 P
/2
g _.
Sta.1
~:* = 0.36; B* . Q.764 8* = 1.34 Fj!+
. l,yl 5* n 2.67
-r,e,=ar - 0.343 e - O.mo 9 = 0.5$4 e = 0.73
2.Z H = 2.68 H u 3.34
6 H - 3.66

y, in.

I 4
I

00 1 I /+464
.2 4 .b.8 /! -u 0. f.4.6.8M
I I 1 I I I u I I 1 I 1 1
U .Z .4 .6 .8 kO olFf.&.8Lo

) I I I 1 I
0 .f ,4 .G .8 40

(~) ~ = 0.96.

Figme 10.- Boundary-l~er velociw profiles along diffus~ and at t~~lpipe

etit for % percent of diffuser le~h roughened (configuration V).

I
 un)

/(6 I I I 1 I I 1 , I sta. 6
I
.ma.5

Sta. 3
B*, ~ ..
/2
$ -
Sta.1 ---

,g+ = 0.154
8 - 9 = 0.124
H = 1.24
y, In.

I N
00 .2 4 O ,2 .4 ,(2 ,8 [0
O ,2 ,4 ,.4.8 /0
~~
0.2:4 .~.8/00,2d,G,8L0
O-D

(b) : = o.g4.

Fltgure 10. - Continued.

!3

,
 I I
r I I I

I eta,
.9
I
ma.6
8ta. 4
t st9. 3
ma. 2 -- A ?
./z
m. 1
-
0.333 8* - 0.932
8* = 1.37 0+ = 2.~o 6* = 1.82.
O.z!lo 0 = 0,372 e = 0.760 L9 m 1*41
8* . 0.148 l.% H - 2.6h H = 2.78 H - 3.51 H - 1.3
8 e - 0.121
H = 1.22
I y, in.
I i I
4

(c) * = 0.915.

I
Fi@re 10. - Concluded.

t
,, ,,,1
~, ..: l. : ,! ?.: ,. , ;,.

1
1: ,.
.,,. .. .. .. . . . .. . ,.). t,t; .,, ,, ...,.,
,., ,,.
1 ... . ..
 7+ w TN 3066

4
y, in.

.6 .8 /. o

(a) 32 percent of diffuser length roughened (configuration VIII).

Figure ~.- Boundary-layer profiles 120 apart on circumference at diffuser

tit tation 6 or %/po 094
 35

Station 6 profile
6 plotted in figure 8(b) >

4
y, in.

/
.,.
.i!!. .

,0s

.0.

o
0 2 .4 .6 .0

(b) ~ percent of diffuser length roughened (configuration VII).

Figure ,11.
- Continued.

. .____ .._ . .
36 WA TN 3066

..
9

7 stat
ion 6 profile
plotted in figure 9(b)

4
YJ in.

o
o .Z 4 .* 6 .0 Lo .

(c) 70 percent of diffuser length roughened (configurationVI).

Figure 11.- Continued.

 37

9
..

7 (

Station 6 profile
plotted in figure 10(b)
6

4
y, in.
/
3

o
o -z .4 .6 .8 Lo
u

percent of diffuser length roughened (confi~ation V).

Figure il.- Concluded.

- .. -
 .
NACA ~ 3066

.2

7
6*, in.

.1
e, w
(a) StatIon1.
I I
o
o 6*

e
.4
~*,1*.
(3
.2
El
e, ti.
(b)Station2.
o

.U
Q
&*,fi .
4 II ~ El
e, ti.
(c) Station3.
0. I I

(d) Station4.

.
El ~
Note: S.qaratedflow

;98 .97 .96 .95 .94 .93 .92 .9/

P1/Po

Figure 12.- %undsry-layer displacement and momentum thiclmesses plotted

against inlet pressure ratio for 52 percent of diffuser length rou@ened
(configuration VIII) .
 39

(e) Station5.
8

U +El
Note: &wrated flow

0=

(f) station6.
Note: Sepmated flow
m
.
El-

I I

o
.96 .97 .96 .95
94
93 .92 .9}
P1/Po

Figure 12.- Concluded.

. .
 40 WA TN 3066

.2

Fj*,fi.
./
e, in.

o
0 6*

13e
4
B*, ~. 0 Q

.2! . I I.
e, in. (b) station2.

o
I I

.)
.

8*, fi.

.4 n
c1
e, in.
(c) station 3.
I I
0

[6

6*, in.
.8
e, h.
(d) StatIonk. .
n I I

:98 .97 .96 .95 .94 .93 .92! .9 I

P1/P()

Figtme 13. - Bmndary-layer displacement and momentum thiclmesses plotted

against inlet pressure ratio for ~ percent of diffuser length roughened
(configuration~).
P
NACA TN 3066
41

_
( (
I
(e) Station 5.

[ .
-1--
.

(f) Station 6.

/ r., 1
El
9, in.

z
5*,~.

/
e,in.
3=$=(g) Station7.

o
.9 5 .97 .96 .95 94 .93 .92 .91
PIIPO

Figure 13.- Concl@ed.

..-_. ___ ______ .- ._. = .

 .2

in.
-/
e, in.
=3=1=
(a) Station1-
()
1

o
0 6*

e
.4

&, in.
.2
e, in.

.%
~ I
in. c) -0

.4 n n
in.
(c) station3.
I 1
o

1.6
.. /.
.
6*,~. @
.8
9, In.
(d) Station4.
I I
0
.98 .97 .96 .95 .94 .93 .92 .91
P1/Po

Figure 14.. Boundary-layer displacement and momentum thiclmesses plotted

against inlet pressure ratio for 70 percent of diffuser length roughened
(configuration VI) .
NACA TN 3066 43

3
0

z
(f) Station 6.

El

a=
2
fj*,
fio
/
f3, in. (g) StationT.

o
.98 .97 .96 .95 .94 .93 .92 .9/
PIIP()

Figure 14. - Concluded.

---...-. __ _______ .. .. . ..__ ..

 44 NACA TN 3066

.
.2

&+, in.
.1
e, in.

.0
0 6*

e
4 0 a
6*, ~.

.2 F1 m
e, in. (b) Station2.
I I
o

.8 .

~*, ~~.
+
e, in.

/!6

E*, ~.
.6
e, in.

o
.9 3 .97 .96 .95 .94 .93 .92 .9/
P1/Po

Figure 15. - Boundary-@er displacement and momentum thicknesses Plotted

against inlet pressure ratio for % percent of diffuser length-roughened
(configuration V).
 3

z
in.
(e) Station5.
1
/
in.

/?
in.

in.

(g) Station7.

o
.98 .97 .96 .95 .94 .93 .92 .9/
P1/Po

Figure 15.- Concluded.

._ ._. .
-.. . . .
 46 WATN 3066

2-0
H Station1

1.0

z.o .
m \
H o u o
Station2
/.o
I

3.0
H

/2.0
so

4.0
Separatedflow
H

3.0
60 I , ,
Note: Sqerated flow
Station >
H
o
50 @
4.0 I I I
station6 Note: Separatedflow
H c I

3.0
2.0- 1
Station7
H

Lo
.
.90 .97 .96 .95 .!34 .93 .92 .9 J
P~/Po

(a) 32 percent of diffuser length roughened (configurationVIII).

Figure 16.- Variation of boundary-layer shape parameter with inlet pressure

ratio.
 47

2.0
Station1
H \
n n
u <
Lo
20
H o
Station2
1.0

3.0
H
Station3
2.0 6 I
3.0 I I .
o 0 0
H
Station4
z.o I

4.0 I
Station5
H
n o
30 H

2,0
Station7
H

Lo
.98 .97 .96 .95 .94 -93 .92 .9/
Plh()

(b) 54 of diffuser length roughened (configurationVII).

l?i~e 16.- cont~ued.

-. . .
4$ UA ~ 3066
.

z.o
Station 1 ,
H

1.0

2.0
H y
Station2
/. o I
3.0 I
HO Station3
0 I
Y
2.0 o
3.0
H
o

2.0
4.0 I
Stition5
.
3.0 0 ,4

2.0
3.0 ~ \
. I I
u
H
station6
I
2.0
>
2.0 I

H Station7
. u
In r
.9? .96 .95 , .94 .93 .92 .9 I
P1/P()

(c) 70 percent of diffuser length roughened (configuration VI) .

Figure 16.-Continued.
7P

NACA TN 3066

2.0 I

Station 1
H
, n.
/.o ~
2.0
0 I
H
Station 2
/.0 1
3.0 I
H Station 3

0 I
2.0 o
3.0 1
H
Station 4
2.0
4.0 I
Station 5
H

3.0 r

4.0
H @ u .

Station 6
3.0 b

2.0 ; ,
Station7
H

10
.- -
.98 .97 .96 .95 .94 .93 .92 .9/
P1/Po

(d) 86 percent of diffuser length roughened (configuration V).

Figure 16.- Concluded.

.-. .. - -. .
 I I
Section of ,
diffuser WS1l ~

z
in. VII Config. V, VII
\__ ._ Config.VI
VI
/
v
~-
--- ./ Separatedflow
I
o
o 4 8 /z Zo Z4 Z8
x, in.

(a) Displacement thickness.

/.0

.8

.6
9, in.

.4
VIZI

.Z
.. /
t- I I Config . -l
WI +
++I+wH+ Sepamted flow

o I 1
0 4 6 /Z /6 fo 24 28
x,in.
(b) Momentum thickness.

Figure 17. - Variation of boundary-layer parameters with distance along

diffuser. P~/Po = 0.95.
 6 I
Confi.g . VIII -

4
H

.
I
3 ,

~i- Config. V

I
4? I

. tinfig. ~
. ...
,...
.
........+ +w+w+ Seprated flow
I
/
O 4 8 /2 /6 20 24 28
x, in.

(c) Emnaary-1.qwr Shqe paramter ,

Figure 17.- Concluded.

u
) P
 \

/.04 \

/
/.03

& h
/,02 . .,
Cmf@uxation

10/
v; $ ;:$Z
VII 1 0.5503
VIII ,0 O.gym
/00

.99

.98
0 4
x, in.

Figure ~8.. Vexiation of ratio of longitudinal static presfiure to

inlet static pressure with distance along longitudinal axiE for
all conflguratiom.
 NAC!ATN 3066 53

.9 I I
Tailpipeexit
7.% e
u
.8
.047L
,-1.
.274L
~ ~1
7 4

4 L
Pideal Diffuserexit
1 .
.6 3 iJ-~ c n

Smooth surface (reference3)

.5
.-
I 1 1 ,
.99 .98 ,97 .% .95 .94 *93 .9Z .9/
P1/Po

(a) 32 percent of diffuser length roughened (configuration VIII).

.9

.6

*7
4
pIdeal
. . . .
.6
I n
El
Diffuserexit
. -1 I
,5
.99 ,9(9 97 ,96
95 .94 .93 .9Z .9/
PJP()

(b) %percent of diffuser lengthroughened (configuration VII).

Figure 19.- Variation of diffuser effectiveness with inlet pressure ratio

for rough-surface configurations.

.. .. ._. . - .c -
 NACA TN 3066

.9
!l?ailpixexit
.

.6

.7
~
41deal
. , .. . .
.6
c 0
+ Diffuserexit

I I

.99 ,98 .97 .96 .95 .94 .93 .92 .9/

P1/Pfj

(c) 70 percent of length roughened (configuration VI).

.

.9

.8 /

F
.7
Ap
*ideal
.
uA
,6
A
Lb LA aA A
Diffuserexit

-1
:99 .98 .97 .96 .95 .94 .93 .92 .9/
P1/P()

(d) 86 percent of diffuser length roughened (configuration V).

Figure 19. - Concluded.

NACA TN 3066 55

./b I
lailpi~exit

n
./2 - Diffuserexit ,% y ~
b
G
I 0 - -- - -

Snmoth diffusertailpipeexit (reference3)

.W -

.99 98 .97 .96 .95 .94 .93 .92 9/

PJPcl
(a) 32 percentof diffu~~rl~@h ro@en~d (~onfi~rationVIII).
,/6 I
milpip exit
Diffuserexit~ r
Q c~
. \
./2 2 E ~ ~
E . . . .
qcl

.B
.99 .96 .97 .96 .95 .94 .93 .9Z .9/
P1/P(J
(b) 54 percentof diffuserlengthrou@ened (configuration
VII).

I6 I
Tailpipeexit

& -

. 0? Wu
.99 .98 .97 .9b .95 .94 .93 .9Z .9/
PIIP()
.- (c) 70percentof diffuserlengthrou@ened (configuration~).
./(0

Tailpipeexit
*/z - 1 I \
& Diffuserexit .
T
c1 *~Al A .t.
.08 5
.99 .90 .97 .96 .95 .94 .93 . 9Z .$7
P1/P()
(d) 86 percentof diffuserlengthro@ened (configuration
V).

Figure 20.- Total-pressure-loss coefficient plotted against inlet

pressure ratio.

.-..
 9 I I 1 I

Config. Percent diffuser

length roughened
......... I 97 (ref. 1) +
8
v 86

----- VI 70
7
54

- VIII 32
6 I
/,

5
y, f n.

o
0 .2 .4 .6 .8 /. o

Figure 21.- Rxmdary-layer veloci~ profiles at diffuser exit, station 6,

for all configurations. q/P~ = 0.95.
NA!JATN 3066 57

.9
I Tailpipe exit I

.8
I
)- Reference 3

.7 -Reference +
Ap
Pideal /_ , . -_
---
I
.6
Diffuser exit

.5

..
0 20 40 60 80 /00-
Percent of diffuserlength roughened

(a) Diffus.er
effectiveness.

.20

/ ---- Tailpipe exit

./6 /
/ \
&- / \
\.
q \

.12

Diffuser exit
I I I I \
.08
o 20 40 40 80 /00
Percent of diffuser length roughened

- (b) Total-pressure-loss coefficient.

Figure 22.- Variation of performance parameters with extent of surface

roughness for ~/pO = 0.95.

NACA-hn@ey-1-22-54-1000
- .
. -. - .-