K ,', '.
October 1941 as
Advance Restricted Report
DEVELOPMENT OF COWLING FOR LCHG-NOSE AIR-COOLED
BMGINE IN THE NACA FULL-SCALE WIND TWNEL
By Abe Silverstein and Eugene R. Guryansky
Langley Memorial Aeronautical Laboratory
Langley Field, Va.
UNIVERSITY OF FLORIDA
120 MARSTON SCIENCE LIBRARY
P.O. BOX 117011
GAINESVILLE, FL 32611-7011 USA
*; NACA WARTIME REPORTS are reprints of papers originally Issued to provide rapid distribution of
dance research results to an authorized group requiring them for the war effort. They were pre-
; lusly held under a security status but are now unclassified. Some of these reports were not tech-
Sedited. All hate been reproduced without change in order to expedite general distribution.
:.. ,... .F
iHv,.1; t l -..* ;... w,",
NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS
DEVELOPMENT OF COWLING FOR LONG-NOS3 AIR-COOLED
ENGINE IN THE KACA FULL-SCALE WI:rD TU:TNEL
By Abe Silverstein and Euigene R. Guryansky
An investigation of cowli:.gs for long-nose radial
engines has teen made on the CO.rtiss X7-42 airplane in
tne FACL full-scale wind tunnel. Tihe XF-42 airplane is
provide- with a Fratt & .Thitneyv R-160-31 eni.-.e, which
has a p'roreller s-iaft aidl tearinr.g .o-c smg that is 2C
inches lcnQ.er than the sta:.d-.r sniort-nose engine cf the
same series. Tr.is forward txz.ension of the rropeller en-
ables the use of fuselage rn,6e sna-es of higher fineness
ratio than are possible rith the olunter snort-nose en-
giae. In ths original C-'rtiss GaT.an.y design of the
XP-42 airplane the pointed fi.sela.:e nose Was used (fig.
1) and sharp-edge Zsco0v0 T.er- added at tce botto- and top
of the corlin6 for the engine-cooiiLg and tae carburetor-
air inlets. Flight tests shored L.he hi-h seed of the
airplane to be co.naprable ritL, but Lot sru-erior to, that
of the P-36, which is a aimiltr airplane -ith a short-
nose en6r'ne and a conve.,tionnl Yia"A co living installation,.
Inspection of the corlin. sc.onps disclosed so".,rces of
dra&, t ie existence of which were substantiated byr pre-
liminar: rIAC-A flight measurements. These tests shored
that the engine cooling air entered the lo-er scnoo at
about half the airtrlane flight velocity and that the
kinetic energy of this flow -as d-ssir-ltEd by the sharm
change in the air-flow direction at the rear of the scoop
and by the expansion fror. the small scoop area to large
area ahead of the er.gi.ie. ,See fig. 2.)
The existence of a large interral er.ergy loss due to
the cooling-air flow was establis.-ed and experience led
to the belief tnat a f',rt.er si-bscsntial external drag
world be added b- the flow over tae sharp scoor e-ges.
The full-'sca-e tu..nel invest i.gation was tien instigated
for the rurn, ose of improving: the original sc.Jop co-ling
or developing an efficient ccwl of another type.
The wind-tunnel program included an initial investi-
: nation of the original p-42 cowling, which was followed
by tests of several modified arrangements with improved
scoops. The general rusatisfactory aerodynamic charac-
teristics of all the cowlings with scoop inlets led to
the develo-rment of the annular high-velocity inlet cowl-
ing. Since it ras the p"-pose of the wind-tunnel i3.vesti-
Latioi. to &evelop an orti:.aim cowling that could be later
conrstri'cted for fli..:ht tests, the various co-linr param-
eters, suci. as inlet velocity ratio, exit area, etc.,
were studied in considerable detail. This cowling has
been constructed and is ci'rrsntIly undergoing flight tests
on the P-42 airplane. Tae results of the flight investi-
gation will be reported at a later date.
METHOD AND APPARATUS
The NACA full-scale w.nd-tunnel balance equipment
used for tnE force measurer.ents is described in reference
1. Tne method of mountiaf the airplane o- the balance is
shown in figure 1. Tihe special tecaiiique end apparatus
used for the momentum measurements are described in ref-
erence 2. Static-pressure measurements rere obtained
either by the use of static orifices or 1/18-inch dian-
eter static-Creszture tubes mo'inled near the airrnlane sur-
faces. The air flo?: tnro.tgh the engine cowling ras meas-
r.red by total-opressure and. static-prersure tubes placed
in te.- dift'.sers ahead of the engine baffles, and in the
RESULTS ANTD DISCUSSIOrS
Lriinael XFr-42 cowling.- A photograph of the instel-
lation of t,.e original scoop co-ling on t.ie XI-42 airplane
is sho!rn in figure 1; a sketch with a more detailed view
of the engine air scoop is c'ro'n in figure 2. Tae cool-
in& air is turnAd -ih a snocrt radius throrvh 900 and dis-
charged. into the coT.mpartn:.nt anead of the enLgine cylinders.
As a result of the energy; losses occurring in the turn and
the expansion, th.e total pressure at the front of the
e.gi.ie baff les was fund to be only about 0.4 the free-
stream total pressure. Tn-.is large inlet loss was chiefly
responsible for tIhe .ifh drag if the cowling installation
and tihe drag coefficient for the airplane equipped with
tre original cooling-air scoop was 0.0040 higher than for
the smooth airplane with the scoop removed and the cowlig.
sealed. Although the internal losses largely accounted
for the drag of the original co-l, a substantial incre-
ment was also added by tae sharp scoop edges. The drag
coefficient for the airplane in the smooth condition (fig.
3) served as a base value for determininin the drags of
all the modifications tested.
Original cowling with multiple scoons.- In order to
avoid the large internal cowl losses, the single original
sharp-edge scoop was replaced with four smaller rounded
inlet scoops (fig. 4). T-e use of milticle scoops rather
than a single scoop was advantageous botn in obtaining
better diffuser passeres a.d in avoiding the snarp oend
required in the single-scooT arrangement. A sketch show-
ing the detailed dimensions of the ducts is contained in
figure 5(a). Tue results obtained with tnis arrangemrient,
whica was designated cowl 1, are shown in table I.
The results were unsatisfactory since it was found
that the flow was separating from thje inner wall of the
duct passages and owing to t".5e negative pressures over
the to; of the covl in the cliii1. condition, the flow'
through the urprer scoct wa; reversed. As a result of the
flow breaczdown ii. tnq ducts, the pressure in front of the
engine averaged only about 0.6 the free-strear', rresst're
(fig. 6). The air-flow quantities measured for three
exit area of 37, 84, and Q8 square inches were .,3,70,
8,810, and 10,280 cubic feet per minute, respectively.
The drae coefficient corresrondin& to the 67-square-inch
outlet area was 0.0023. Tue drag of the airplane with
the scoop outlets sealed and with the inlets unsealed was
increased 0.0017 above the drag of the smooth airplane.
As a result of the difficulties encountered with the
four-scoop arrangement, the ton scoop was removed and the
scoop inlets were extended forward along the cowl about
11 inches (see fig. 7); with tnese changes the duct inlet
area was considerably reduced. (See fie. 5(b).) The
modifications served to locate the inlet more nearly
normal to the local floor" direction and to lengthen the
diffusing passage. Tie results were somewhat more satis-
factory and the total pressures in front of the engine
were higher than for the former arrangement. (See fig.
The diffuser passage, however, was still inefficient,
since flovi breakaway occurred on its inner wall and a fur-
ther modification was made in which the duct passages were
straightened. (See fig. 5(c).) For this improved arrange-
ment, with an outlet area of 91 square inches, a drag coef-
ficient increment of 0.0024 was measured with nn air-flow
quantity of 12,700 cubic feet per minute. With the scoop
inlets and outlets sealed, the airplane drag coefficient
was increased 0.0006, which is a measure of the effect of
the protruding scoops on the external drag of the airplane.
Annular inlet corl.- Since the drag of all the scoop
arrangeme.-ts tested was high, the investigation was di-
rected toward developing a cowl in which the cooling air
was introduced through a narrow annular inlet at the nose
of the airplane, with a spinner fairir.g for the propeller
hub and the blade shanks (figs. 8 and 9). This cowl,
which is designated cowl 2 (taole I), was designed so that
the velocity at the coolin6-air inlet was about one-fourth
of the free-stream velocity. It was tested first with the
exit sealed and the airplane drag was increased 0.0012,
owing to the cowl. form drag and the circulation of air in
the cowl opening. With the inlet also sealed, the air-
plans drag was increased by only C.0003. For different
outlet areas, the airplane drag coefficient was increased
0.0022 with an nir flow of 12,050 cubic feet rer minute
and was increased 0.0027 *-ith an air flow of 17,000 cubic
feet per minute.
These drag increments caused by air flow were too
large and since dra6 reductions that should have been ex-
rected cwing to improved internal flo- (see fig. 6) were
not fully realized, it was suspected that the cowl outlet
was unsatisfactory. Tuft investigation of the flow from
the original cowl outlet on the XP-42 airplane (fig. 10)
showed that air was bir.-A discharged over and around the
exhaust collector and flap gear in such a manner that the
flow over the fuselage was disturbed. The outlet was
modified, as shown in figure 10, by removing the conven-
tionl?! flap gear and exhaust collector from behind the
engine and installing a smooth unrestricted outlet. Tith
this modification, tne drag coefficient was reduced 0.007
and the cowl drag coefficient of 0.0015 was measured with
a flow of 12,040 cubic feet per minute. The investigation
was continued by sealing the conventional radial cowling
outlet and providing a bottom outlet on the cowl. (See
fig. 11(a).) This bottom outlet was too small because
the measured air flotr was lover than required and a larger
bottom outlet ras constr-.cted (fig. 11(bj). The corl drag
coefficient for this arrangement was 0.0011 -ith air flow
of 12,800 cubic feet per minute. This drag is 0.0004
lower tnan for the cowling with the smooth radial outlet
and is 0.C0011 lower than the conventional flap outlet.
The large drag reductions effected with the improved
outlets emphasize the imrortance of providing a smooth
outlet on production airplanes. Although the single bot-
tom outlet will probably be insufficient to provide uni-
form cooling for all the engine cylinders, the result ob-
tained with this arrangement is of particular interest as
a reference for evaluating the erag of the outlets.
From pressure measurements in the diffuser of the
annular cowl 2 (fig. 6), it was noted that tae total pres-
sure was less titan 0.9 the free-stream dynamic pressure.
Since it was expected tnat tnis value would be close to
stream pressure, the flo-7 over the spinner was investi-
gated with tufts. It was foune that floor reversal was
occurring on the unper Fpart of the spinner a. the inlet.
This phenomenon was furtuear investigated by measurements
of pressures along the spinner, wnich are snorn in fig-
ures 12 and 13. In tnesa figures the rmagnitude of the
pressure is indicated as the length of the vector normal
to the spinner surface. It w-ill be noted that a large
adverse pressure gradieit exists in the direction of air
flow, the value of which is indicated by the slope of the
pressure plots. For the climb condition the slope is
high forward or. tae spinner and shows a jagged peak ahead
of the cowl inlet. For the high-speed lift coefficient
(CL = 0.150) the adverse pressure gradient is high toward
the forward part of the spinner and decreases several
inches ahead of the nose of the inlet. In agreement with
usual boundary-layer phenomena, the extent of tuft rever-
sal could be coordinated .with the slope of the pressure
gradient along the spinner. Further modification wes then
made to cowl 2 (fig. 14) to reduce the pressure gradient
along the spinner. The inlet area for the cowling was re-
duced by increasing the spinner size (spinner B, fig. 9)
so that the inlet-velocity ratio (Vi/V) was increased
above 0.5. With the higher inlet velocities, the diffuser
pressures were increased to approximately 0.97%o. The
pressures on the spinner corresponding to the two outlet
conditions tested are shown in figures 15 and 1'6.
Tor the high-speed condition the rise in pressure
along the modified spinner is considerably lower than for
the original.spinner. In the climb condition, the same
irregularities in the pressure distribution occurred and
these irregularities were found to be associated with a
tuft reversal even for the higher inlet velocity obtained
with the spinner modification. The pressures at the face
of tLe engine in the climb condition were, however, uni-
form and high.
The drag coefficient for cowl Z and spinner B, with
the modified bottom outlet, was 0.0006 -rith an air flow
of 13,870 cubic feet cer minute. The gain due to increas-
ing the inlet velocity and recovering the full total ires-
sure in the diffuser amounted to 0.0005. The drag coef-
ficient of 0.0006, meas-,.c- ca for this arrangement, is the
lowest that aas been obtained in full-scale tunnel tests
on radial-air-cooled engine cowlings. The efficiency of
the cowl is most clearly demonstrated by the total-pres-
sure measurements at the outlet (fig. 17). By progres-
sive modifications the outlet total pressure was in-
creased from an average of about 0.3qo with cowl 1, to
more than 0.8qo with cowl 2 and swinner B. Since the
internal drag losses are a direct function of the factor
1 increasing the value of H/qo at the outlet
from 0.3 to 0.8 corresponds to reducing the internal drag
losses to almost one-fifth.
In order -to determine whether the high efficiency of
this cowl could be preserved with greater air flows, the
outlet area was increased about 50 percent by partly open-
ing the smooth radial outlet. A cowl drag coefficient of
0.0012 was measured with an air flow of 21,140 cubic feet
per minute. This air flow is sufficient not only for the
engine air but also for the carburetor and oil cooler.
An investigation of ducts for handling the oil and the car-
buretor air was not made.
In order to aid in the estimation of the compressi-
bility effects and to study tne flow within the annular
diffuser passages, pressure measurements were made over
the inside .and the outside of the cowl for several dif-
ferent air-flow conditions. These plots are shown in fig-
ures 18 to. 2.1. The maximum negative pressure of approxi-
mately 0.4q was measured at the nose of the cowl, which
indicates that the critical compressibility speed will
occrr above 500 miles per hour at 20,000 feet altitude.
The uniform recovery pressure on the inside of the duct
is demonstrated in figures 19 and 21.
1. The long-nose engine enables tnre design of an
efficient annular inlet corlinG owing to the length avail-
able for a diffusing passage.
2. Tie ratio of the cool-rLe-ai.r velocity at the
cowling irnlet to the itre.am velocit:' is one of the most
imnortanL design variables for tne anr.ular inlet co-ling
and this ratio should not tie less tnan about 0.5.
3. The critical co..,rresz utility speed for the long-
nose engine cowling can be extended to above 500 miles
per hour at 20,000 feet altitude.
4. Important draz losses occur du3 to the flow of
cooling air out of conventional cowlinE outlets with flap
gear and erhaust collectors to disturb the flow.
Langley Memorial Aeronautical Laboratory,
National Advisory Committee for Aeronautics,
Langley Field, Va.
1. DeFrance, Smith J.: The E.A.C.A. Full-Scale ,Tind
Tunnel. Rep. 1o. 459, :TAC,, 1933.
2. Goett, Earry J.: Experimental Iivestigation of the
komex.tum !ethod for Betermining Profile Drag. Rer.
Yo. 660, UIACA, 1939.
NACrA Table I
TABLE 1. SUMMARY OF RESULTS
ofnn a eDrag coefficient
Cool:rng aoteo tao ZOO i n)
Exit (at 100 mp Air quantity Inlet
Width area 6 C0 at (Wu ft per velocity
Oowl Sketch outlet Test conditions e In. I C at CL=O mln at ratio
opening D D 350 mph) 71
(in.) CL=-015 (b) -_
without Sealed 0.0192 0.0203
with 1.49 Itandard 167 0.0232 0.0243 0.00o0 16,100 0.69
CowJ 1 Sealed 0.0209 0.0220 0.0017
5/8 67 .0212 .0226 .0023 6,970 0.15
3/1 84 8, sio .19
7/8 98 10,280 .23
Sealed )11 cooler open .0210 .0224 .0021
Cowl L Sealed Scoops sealed 0.0194 0.0209 0.0006
modifie5/8 .0196 .0209 .0006
5 5'S coops open 63 .0206 .0221 .0018 7,330 0237
/I /4 .0208 .0224 .0021
7/8 98 .0210 .0225 .0022 10,900 .34
ut sanonghduened. 65 .0211 .0224 .0021 9,160 .28
7/8 Same as 5/8 91 .021b .0227 .0024 12,700 .39
Cowl P Sealed 0.0200 0.0215 0.0012
Spinner 5/8 70 .0213 .0w25 .0022 12,050 0.32
3/i4 78 13.750 .36
7/8 98 .0216 .0230 .0027 17,000 .414
Sealed Oil cooler open .0209 .0222 .0019
5/8 (e) 63 .0204 .0218 .0015 12,040 .31
Sealed toae sealed a .0192 .0206 .000'
Bottom exit open a 72 .0198 .0214 .0011 9,940 .26
Modified bottom exita 91 .0199 .0214 .0011 12,800 .53
u Modified bottom exit 91 .0199 .0212 .0009 13,550 -35
upper Inlet sealed
Partial 5/8 Modified bottom exit 136 .0202 .0216 .0013
5/8 Bottom sealed a 45 8,150 .21
7/8 a 63 12,100 .32
1-1/4 a 90o .0209 .0224 .0021 18,600 .49
Oowl 2 0 Sealed Modified bottom a 91 0.0199 0.0209 0.0006 13,870 0.55
Spinner Partial5/8 *" a 131 .0204 .0215 .0012 21,140 .83
Cowl flap gear removed and smooth exit-Inetalled.
b Based on smooth condition with original scoop offi landing gear fairing removed; control surfaces unsealed;
and antenna on.
Figure 1.- The XP-42 airplane in the standard condition,
Figure 14.- Tie XP-42 airplane in the sccth condition with
cowl 2 modified and smooth cowl flaps.
NACA Figs. 2,10
I N ] '-'
-.--- /' 1^ I
( i h. 'I K / K
I," N 1
*~~ 'J 'A'
I .- ; : '
The XP-.'2 airplane in the completely smooth condition
mo-nted in the full-scale tunnel.
AiLture 4.- The XP-42 airplane in th7 sr-coth condition with cowl 1
and original cowl flaps.
La i-e0 on to- L -a
0 '1.'" 0 tu. Jad S0 ,OI .I .
CU, It L->"/ f Ls3id u icf- o i ^ .
I -r --
Figire 7.- The XP-42 airplane in the amcoth condition with cowl 1
modified and criminal cowl flaps.
FiLuire S.- The XP-42 airplane in the succth condition with ccvl 2,
spinner A, and original ccwl flaps.
.* Q ..
S 4 r ..- .,
t ------- ..
(a) Bottom outlet
(b) Modified bottom outlet
Figure 11.- Cowl outlet on XP-42 airplane.
NA : 4,
2 4_^? __
tre I. es re. dirir a. over ,op of sponrier A w/, n in'let ve.-.c, t r r' .. I .
2 4 6 .3 i.r
A ~titF*/e j Pre-ss -,j e .,-. '.. ,.'4 u'' ..i i f w i/f *i I y er-ci rf n e, a!;.
.2 g L JO
war. J$ Prr-.,ac~re ~ .tr ~ cc are, 'op os %c.'lrrr .S .4.0) r:et vcA'tr rd1 ,1$rsS
.2 4 6 t- a
fart pra- r*-e o .' tQ' over tap ct J'-n.ner B ith n e ..dtoC' t 't2iS.
- ,-, .
-. 0 _.
II 3, .-
I J%, *,;
l f ^
'/t~fi^ J~L'. T.*/? ---
UNIVERSITY OF FLORIDA
3I 60I10illlll ll I 4 llllU ll
3 1262 08106 479 1
UNIVERSITY OF FLORIDA
120 MARSTON SCIENCE LIBRARY
P.O. BOX 117011
GAiNESVILLE, FL 32611-7011 USA