An experimental investigation of the aerodynamics and cooling of a horizontally-opposed air-cooled aircraft engine installation

NASA Contractor Report 3405

An Experimental Investigation of

the Aerodynamics and Cooling of a

Horizontally-Opposed Air-Cooled

Aircraft Engine Installation

Stan J. Miley and Ernest J. Cross, Jr.

Texas A&M University College Station, Texas John K. Owens Mississippi State University Mississippi State, Mississippi David L. Lawrence Turbo West Corporate Aircraft Center Broomfield, Colorado Prepared for Langley Research Center under Grant NSG-1083

N/LSA

Nationa_ Aeronautics and Space Administration Scientific and Technical Information Branch ,1_ -¸ SUMMARY A flight test based research program was performed to investigate the aerodynamics and cooling of a horizontally- opposed aircraft engine installation. Specific areas inves- tigated were the internal aerodynamics and cooling mechanics of the installation, inlet aerodynamics, and exit aero- dynamics. The applicable theory and current state-of-the- art are discussed for each area. Flight test and ground test techniques for the development of the cooling installation and the solution of cooling problems are presented.

The results show that much of the internal aerodynamics and cooling technology developed for radial engines are applicable to horizontally-opposed engines. Correlation is established between engine manufacturer's cooling design data and flight measurements of the particular installation. Also, a flight test method for the development of cooling require- ments in terms of easily measurable parameters is presented.

The impact of inlet and exit design on cooling and cooling drag is shown to be of major significance.

INTRODUCTION The research program, which is reported herein, was established to perform an exploratory investigation of the cooling drag associated with reciprocating engine powered

general aviation aircraft. As work progressed, it became

apparent that attention should be focused on the engine

cooling installation aerodynamics which are manifested as

cooling drag. An associated area of concern, inadequate

engine cooling, particularly for supercharged engines at

altitude, is also related to installation aerodynamics.

Poor aerodynamic design not only results in excessive cool-

ing drag, but also can result in poor cooling. The standard

cure for inadequate cooling is to operate the engine at

higher-than-necessary fuel flows. Consequently, both

situations result in reduced fuel efficiency.

Cooling installation aerodynamics is concerned with the

behavior of the airflow system required for cooling by air-

cooled reciprocating aircraft engines. The components which

make up the system are dependent on tbe particular engine

geometry. Figure 1 illustrates four basic engine geometries

and the associated cooling airflow system configurations.

Three of the four geometries, the in-line, the vee, and the

horizontally-opposed, use the same system, consisting of an

inlet, high pressure plenum, low pressure plenum, and exit.

The first three components are necessary because the engine

geometry requires the cooling airflow to pass through the

engine perpendicular to the flight path. The radial engine

geometry, however, requires a relatively simpler system

consisting primarily of a cylindrical cowl. The airflow

passing through the engine remains parallel to the flight

path. The cowl functions as an external baffle, forcing the

captured airflow through the coolin_ fins. It also in-

corporates the cooling flow exit in its structure.

Cooling installation aerodynamics involve two problem

areas. The first is concerned with the character of the

flow through the engine's cooling fins and the resulting

heat transfer to this flow. The technology developed for

radial engines in this area is also applicable to horizontal-

ly-opposed engines as well as in-line and vee geometries.

The second problem area is concerned with the external and

internal flow of the installation system. The existing

technology here is applicable only to similar geometries.

The large amount of data regarding radial engine cowl/nacelle

design is of little use to horizontally-opposed installations.

As part of the research program, an extensive literature

survey was performed to identify material applicable to the

cooling flow/heat transfer aerodynamics and cowl external

internal aerodynamics problem areas. Over five hundred

references have been collected. The results of the literature

survey will be published as a separate report.

The literature is dominated by the development of radial

engines and mainly encompasses the period 1930-1945. After

this time, interest turned to the development of gas turbine

powerplants, and air-cooled engine related technology essen-

tially disappeared from print. All aspects of the problem

areas are well covered except for the cowl aerodynamics of

the non-radial _eometries. Of the five hundred plus cita-

tions during the radial development period, only six (ref-

erences 1-6) concern in-line or vee cowl aerodynamics.

These are, for the most part, also applicable to the

horizontally-opposed cowl geometry. In recent years,

attention has been given to horizontally-opposed installa-

tions (references 7-11).

Except for the literature survey, the research program

was experimental and involved a series of flight test

investigations. One ground test investigation on the test

aircraft was also conducted. The first test program attempt-

ed to measure the cooling drag of a single engine aircraft.

The remaining programs investigated various aspects of the

installation aerodynamics utilizin_ a twin engine aircraft.

The investigators wish to acknowledge the important con- tributions by the general aviation industry to this research program. In particular, the donation of the aircraft and propulsion system for use in this program by Piper Aircraft, Avco Lycoming, and Hartzell Propeller is greatly appreciated.

Also, the participation by engineering representatives from Avco Lycoming, Beech, Cessna, Grumman American, Mooney, Piper, Rockwell, and Teledyne Continental for program critique and review was an important asset.

SYMBOLS

a effective orifice area, m 2 a coefficients of least squares surface used in 1,2,3 equation (5) b exponent of orifice power law C constants of heat transfer power laws used in 1,2,3,_ equations (8) - (II) pressure coefficient P - P= Cp qco H heat transferred per unit time, joule/sec I indicated engine power, kw k,l,m,n exponents of power laws in development of coolin_ correlation relation, equations (8) - (13) P static pressure, N/m 2 P_ free stream static pressure, N/m 2 CO free stream dynamic pressure, N/m 2 T ambient (free stream) temperature, °C a exhaust gas temperature, °C TEGT T temperature of cooling flow exiting the ex engine, °C T effective combustion gas temperature, °C g cylinder bead temperature, °C T h average of the 6 engine cylinder head Th A temperatures, °C cylinder head temperature of cylinder Th 6 number 6, °C

T

temperature of cooling flow in upper plenum,

up

at rear of engine, °C T* heat transfer temperature ratio heat transfer temperature ratio based on use of ThA for cylinder head temperature heat transfer temperature ratio based on use of Th6 for cylinder head temperature Vo inlet velocity, m/sec V flow velocity approaching inlet, m/sec O W cooling air mass flow, kg/sec W charge flow of fuel and air through the C engine, kg/sec coordinates used for integration in the x,y determination of cooling air mass flow, equation (5), m ap engine baffle pressure drop, N/m 2 air density, kg/m _ density ratio relative to standard sea level density ratio of cooling air flow exiting °ex the engine relative to standard sea level MEASUREMENT OF COOLING DRAG The objective of this program was to ascertain if reliable measurements of cooling drag could be obtained using flight test techniques. If so, then follow-on testing would attempt to identify the contributions of specific in- stallation components to the cooling drag.

The aircraft used for this program was a Beech T-34B

(Figure 2) obtained on loan from the U.S. Navy. The cooling

installation of the aircraft is unique by current practice

in that exhaust ejector pumps, or augmentors, are used at

the exit. The ejector tubes are visible in the figure

extending through the bottom of the fuselage, near the

leading edge of the wing. The drag measurements were made

usin_ the technique of feathered sinks, i.e., engine stopped,

propeller feathered, gliding flight. A special full-feather-

ing propeller was installed for this program.

All flights were performed in the early morning and in

calm air. Airspeed deviations for good data runs were kept

to witbin one knot. A stabilized glide of at least 1,500

meters in altitude was required for an acceptable data run.

Configurations which were tested included inlets open, in-

lets closed, ejector tubes open, tubes restricted, and tubes

closed. The results of the drag measurements are given in

Table I.

The drag associated with the cooling flow through the

installation is indicated to be seven percent of the no-flow

aircraft drag. Drag calculations based on the momentum loss

of the internal flow yielded a value of six percent of the

no-flow drag. These values are similar to those of reference

11 for a twin engine configuration.

As a result of this first drag measurement program, it

became evident that the drag levels of interest were barely

resolvable through flight test. The follow-on program to

investigate the drag contributions of specific installation

system components was abandoned in favor of studying the

associated aerodynamics. The point-of-view here is that

cooling drag can be reduced to the minimum necessary level

by good aerodynamic design of the cooling installation. The

emphasis of the research program thus shifted from a tradi-

tional trial-and-error drag clean-up approach, to a more

scientific approach of investigating and understanding the

various aerodynamic effects involved in the operation of the

cooling installation.

COOLING INSTALLATION AERODYNAMICS PROGRAM

Test Aircraft and Cooling Installation

The test aircraft used in the program was a Piper PA-41P

prototype pressurized Aztec. The aircraft is shown in Figure

3. The engines were turbosupercharged with a sea level rating

of 201 kw. The aircraft was capable of operating at altitudes

in excess of 7,000 meters. The starboard power plant was used

for the various studies of the program.

A schematic of the cooling installation of the PA-41P

test aircraft is shown in Figure 4. Particular points of

reference are denoted by numbers. This arrangement is typical

of most horizontally-opposed engine installations. The source

of cooling flow is represented by point (I). The flow has

been acted on by the propeller and is no longer at free

stream conditions. The cooling air enters the inlet (2) and

is conveyed to the upper plenum (3). It then flows through

the cooling fin passages into the lower plenum (4) and is

exhausted through the exits (5). This configuration, where

the cooling air flows from the upper to the lower plenum, is

known as downdraft cooling. Updraft cooling configurations

are also available, where the air flows from bottom to top.

The location of exits is generally on the lower surface of

the cowl, although some arrangements exist with the exits on

the upper surface.

Cooling Installation Operation and Design

Cooling requirements. The function of the cooling installation is to conduct sufficient air flow to cool the engine under specified operating conditions. The airflow requirements and engine orifice characteristics are normally supplied by the engine manufacturer in a form similar to Figure 5. The coolin_ requirements are determined from the right side of the graph. The operation condition is specified in terms of air temperature, engine power, and cylinder head temperature (CHT); the required airflow is then read from the ordinate. The engine orifice characteristics refer to the

relationship between the rate of flow through the cooling fin

passages and the decrease in pressure which accompanies this.

This pressure decrease is called the baffle pressure drop,

which relates to the intercylinder baffle plates used on

most air-cooled engines.

Theory of operation. Figure 6 gives a schematic of the cooling installation model. The source of the cooling air (i) is the free stream with a dynamic pressure head corresponding to the airspeed of the aircraft. In actuality, the pressure head is modified by the propeller. The inlet (2) captures the coolin_ air, partially converts the dynamic head to static pressure, and conveys the flow to the upper plenum. The upper plenum (3) serves as a reservoir for the engine and auxiliary cooling flow. The cooling air at this point should be at stagnation conditions with full recovery of the dynamic head.

The flow then proceeds through the cooling fin passages of the engine into the lower plenum (4). If the upper plenum is used to supply auxiliary cooling (7), then part of the air flow takes this path. In passing from region (3) to region (4), the air is heated and the density changes accordingly. The heated air then accelerates through the exit (5) such that its static pressure is equal to the local external (6) static pressure of the flow. Through the use of a hinged flap, both the exit area and the external exit pressure can be varied to control the flow rate.

I0

In actuality, the pressure recovery in the upper plenum

may be as low as fifty percent of the free stream dynamic

pressure, due to flow losses through the inlet. Also, in-

sufficient plenum volume results in finite velocities, con-

sequently, the flow is not evenly distributed over the engine

face nor is the transition from horizontal flow to vertical

flow through the cooling passages accomplished efficiently.

Cooliq$ installation design. Procedures for the design analysis are given in references 7 and 12. They are summarized here.

The design of the cooling installation is based on one- dimensional subsonic compressible flow theory. Starting with the dynamic pressure head in front of the inlet [point (I) in Figure 6], the upper plenum pressure (3) is determined by applying a pressure recovery factor which accounts for the amount of diffusion developed by the inlet (2) and the flow losses incurred between (I) and (3). There is a question as to whether the plenum static pressure or the total pressure should be used here. This will be addressed in a later section. The lower plenum pressure (4) is determined by the baffle pressure drop associated with the required cooling air flow rate. As indicated in Figure 5, the pressure drop is altitude dependent. The flow density in the lower plenum is determined from the temperature rise across the engine.

This information heretofore has not been supplied by engine II manufacturers, and estimates based on experience must be used.

Typical values range from 50°C to 70°C. The exit area (5)

is then sized to accelerate the flow so that its static

pressure is the same as the local external flow (6). The

exit area acts as the system throttle in that the flow rate

and associated pressure drops will adjust so that the exit

pressure matches the external pressure. For a given flight

condition, expanding the exit area increases the flow rate, and conversely.

The design problem is made difficult by the wide variation

in horizontally-opposed aircraft engine configurations. Of

prime concern here are the upper and lower plenum volumes.

With the oil sump located on the bottom of the engine, in-

stallations tend to have relatively large plenum volumes

below and relatively small plenum volumes on top. The charac-

ter of the flow in these respective regions is affected by

the location and configuration of the induction and exhaust

lines, and the presence of an inter-cooler, alternator, and/or

propeller governor. This leads to some engine dependent

empiricism in the determination of pressure recoveries and

flow losses. Also, the differences between the test configura-

tion, for which the cooling requirements are determined, and

the installation configuration create uncertainties as to the

application of the requirements data. A typical cooling

requirements test configuration is shown in Figure 7. This

is the ideal cooling configuration in that a true plenum

exists on top of the engine and the cooling air is distributed

uniformly across the upper engine face. The temperature of

the cooling air is uniform and the plenum pressure is a true

measure of the baffle pressure drop. For the installation

configuration in Figure 4, the flow is highly nonuniform, the

temperature of the flow rises as it progresses towards the

rear, and the relationship between this plenum pressure and

that of Figure 7 is open to question.

Cooling Installation and Aerodynamics Investigations

The objective of the Installation Aerodynamics Program

was to investigate the various aerodynamic effects involved

in the operation of the cooling installation. This was done

in the context of the aforediscussed design problems in the

previous section. Three areas were studied: internal flow

mechanics, inlet effects, and exit effects. The internal

flow investigation dealt with two problem areas. The first

consisted of an evaluation of the various methods of instal-

lation flow pressure measurement in current use by the gen-

eral aviation industry. The second problem area dealt with

the engine orifice characteristics and the correlation between

installation measurements and test cell measurements. Much

was drawn from the radial engine cooling correlation work

performed by NACA.

The inlet and exit studies investigated the effects of

some basic design parameters on installation performance.

Four different inlet configurations were studied in terms

of both the external flow and internal flow. Exit design

parameters investigated included exit area, location, and

cowl flap geometry.

INTERNAL FLOW STUDIES

Internal Flow Instrumentation Investigation

The objectives of the internal flow instrumentation

investigation were to measure the flow temperature and

pressure distributions and pressure losses through the

installation, and to evaluate different techniques for

measuring the engine baffle pressure drop. Total pressurek

surveys, utilizing Kiel tubes, were taken at several long-

itudinal stations in the high pressure plenum. These

locations, illustrated in Figure 8, were at the rear of

the inlet duct in front of the leading cylinders, and above

each cylinder on its center line. Total pressure Kiel tubes

were also located in the exit ducts. The inlet Kiel tubes

are shown in Figure 9, and the cylinder mounted tubes are

shown in Figure I0. Also shown in Figure I0 are the plenum

temperature probes which consist of a thermocouple sensor

and radiation shield. The temperature probe locations are

given in Figure 8.

The pressure distribution on the upper engine face and

the baffle pressure drop across the engine were measured by

a number of different probes and methods. Representative

techniques of both airframe and engine manufacturers were

included. Figure II illustrates the various probe configu-

rations and locations. All probes shown in Figure ll(a),

except the baffle button probe (I), are 1.6 mm diameter

open-end total pressure tubes. The tube opening was in-

ternally chamfered to a 60 degree included angle. This

increased the probe angularity insensitivity to approximate-

]y 30 degrees. The vertical positions of these probes are

shown in Figure ll(b). The cylinder barrel tubes (2) and

cylinder head tubes (3) were located vertically on the

center line of the cylinder. Cylinder head tubes (4) were

located 9.5 mm below the local fin height on the exhaust

stack side of the cylinder. Cylinder head tubes (5) were

located between adjacent cylinders, flush with the top of

the local fins. Referring to Figure I0, the (5) tubes were

exposed to the engine face pressure without fin passage

losses. The "baffle button" probes (I), Figure ll(c), con-

sist of a brass roJndheadmachine screw inserted through the

intercylinder baffle at the base of barrels. The screw is

drilled for, and fitted with, a 1.6 mm tube for connection

to pressure measuring instrumentation. The head of the screw

is filled and smoothed. Piccolo tubes, Figures 8 and ll(d),

were mounted in the upper and lower plenums to provide an

integrated or averaged measurement of the static pressure.

The upper plenum static pressure was also measured by multi-

element pressure belts. As shown in Figure ll(e), the belts

were attached to the inside upper surface of the cowl. Hole

spacing between belt elements was 5 cm.

The pressure in the lower plenum was measured by four

different probe configurations. Common practice here is to

use total pressure tubes located so that they are shielded

from any local high flow velocities. The total probe con-

figurations used are shown in Figures ll(c) and ll(f). A

set of baffle-shield probes was located in the lower plenum

at each of the baffle button positions (I) in Figure ll(a).

Fin-shield probes were located adjacent to each of the cylin-

der head upper plenum pressure probes (5). All lower plenum

pressure probes, of the same configuration, were manifolded

together to give a single averaged measurement for that

configuration. The fourth probe configuration used was the

aforedescribed piccolo.

The thermocouple temperature probe locations in the upper

plenum are shown in Figure 8. Two additional probes were

positioned in the lower plenum, one at each exit.

The pressure and temperature data were recorded on an

analog tape recorder usin F a serial multiplexing format. A

total of 144 channels of pressure data and 48 channels of

temperature data were available. An 80-tube photomanometer

system was also used for additional pressure data when

required.

The purpose of this study was to evaluate different

probes and techniques for measuring the engine face, upper

plenum and lower plenum pressures. A fundamental problem

involved in these measurements is the question of whether one

is measuring a static or a total pressure. If the plenum

volume is large, then for the range of cooling air flow rates

encountered, the two pressures are the same. If, however, one

side of the engine is tightly cowled and the plenum volume

is correspondingly small, there will be a distinct difference

between static and total pressure. For the PA-41P test air-

craft, the lower plenum was large and the upper plenum was

relatively small. The results should be interpreted accord-

ingly.

Figure 12 presents a comparison of the different lower

plenum pressure measurements. The data represent different

airspeeds, altitudes, and cowl flap settings. All pressures

are referenced to free-stream total pressure. As indicated

in Figure 12, all methods give essentially the same measure.

The fin-shield probes give a reading 3% below the piccolo and

the baffle-shield-down probes give a reading 2% above. The

baffle-shield-up probes read the same as the piccolo. The

differences are felt to be due to position error type effects,

i.e., effects associated with the location and orientation of

the open end of the probe. From the standpoint of simplicity,

the piccolo tube appears to be the best method of measuring

the lower plenum pressure. However, as will be shown, if the

plenum volume were small, consideration should be given to

the baffle-shield-up probes.

The volume of the upper plenum is measurably smaller than

the lower plenum. The cross-sectional area of the upper

plenum is approximately 650 cm 2 , and this combined with the

cooling air flow rate results in flow velocities in the neighborhood of 15 m/sec. There is correspondingly a differ- ence between static and total pressure in the region. Pressure data from the engine face probes and the belts are presented in Figure 13 for two different inlets. The data are given in pressure coefficient form referenced to free-stream static pressure. The abscissa represents the longitudinal coordinate referenced to the cylinder number. The ordinate intersection is the front of the engine; to the right is progressing towards the rear down the right bank of cylinders, and to the left is progressing down the left bank of cylinders.

Several observations can be made regarding the data in Figure 13. All pressure measurement methods were subject to position error effects. This is due to having finite flow velocities of irregular directions in the plenum. The

increased scatter for the climb condition is believed to be

due to the corresponding change in the propeller slipstream

flow. The scatter pattern also varies according to the

different inlets tested. Again, this indicates a change in

the character of the flow through the plenum.

Since the pressure belt measures the static pressure

variations, it is evident that the pressure at the engine

face is also static. There is no effective recovery of the

plenum dynamic pressure. The increase in static pressure

from front to rear is consistent with the diffusion effect

accompanying the progressive passage of the flow through the

engine to the lower plenum. The left side of the plenum

(cylinders 2-4-6) behaves differently from the right side

(cylinders 1-3-5). This asymmetrical behavior showed itself

in a number of different measurements and is believed to be

due to inlet flow blockage by the propeller governor. The

governor is visible inside the inlet shown in Figure 9.

The baffle button probes (I) provide the most reliable

measure of the engine face pressure. The piccolo tube

indicates low. It is possible that the piccolo reading

may be raised through biasing the tube by cutting it short

so that it does not extend to the front cylinders where the

flow velocity is highest.

In summary, results of this investigation show that the

baffle button probe gives the most reliable measure of engine

face pressure. If the engine is not equipped with inter-

cylinder baffles, then a shortened piccolo tube will work

equally well. The pressure in the lower plenum is measured

accurately by the piccolo. However, if this volume is

small, then the baffle-static-up probe configuration should

be considered.

Engine Orifice Characteristics

Background. Engine orifice characteristics relate the cooling air mass flow through the engine's fin passages to the pressure drop across the engine. The relationship is similar to orifice flow, which is described by w = a p_-_p. (I) A more general form of this relationship is given by the power law w = a (pap)b (2) In equation (I), "b" has the value of b = 0.5.

A study of air-cooled engine development shows that the power law relationship of equation (2) is a valid represen- tation of engine orifice characteristics. The constant of proportionality "a" functions as an equivalent orifice area 2O

in that it tends to vary directly with cooling fin spacing

and total passage area. The exponent "b" is a function of

fin spacing and intercylinder baffling. Fin spacing tests

reported in reference 13 show "b" ranging from b = 0.78 for

0.5 mm spacing to b = 0.50 for 5 mm spacing. Available data

for current horizontally-opposed aircraft engines has "b"

ranging from b = 0.52 to b = 0.58. Values of "a" vary over

a much wider range, depending on geometric engine size and

number of cylinders.

The values of "a" and "b" for a particular engine model

are determined by ground test, using a system similar to

that shown in Figure 7. Cooling air mass flows and corre-

sponding baffle pressure drops are measured over the range

of interest. The coefficients "a" and "b" are then deter-

mined by rewriting equation (2) in logarithmic form, (3)

In (w) = b " In (_Ap) + In (a)

and applying linear regression techniques.

Once determined, the coefficients can be used with equation (2), in theory, to measure the cooling air flow through the engine in flight. This approach is necessary because aircraft engine installations do not lend them- selves to direct cooling air mass flow measurements, whereas the corresponding engine baffle pressure drop is readily measureable. For the purpose of flight cooling flow measure- ments then, the engine is used as an orifice meter. The validity of such measurements depend on two considerations: altitude and engine heating effects on the ground test deter- mined orifice characteristics; and correlation between the baffle pressure drop measurements of the cooling flow for the aircraft installation configuration and for the ground test configuration. The first of these concerns will be discussed in the followin_ paragraphs, while the second will be dealt with in a later section.

Altitude and heating influences on engine orifice charac- teristics were investigated as part of the cooling correlation research effort associated with radial air-cooled engine development. The objective of this effort was to extrapolate ground test determined cooling requirements data to operation- al altitudes and different power and mixture settings. Sum- maries of the engine orifice characteristics investigations are given in references 14 and 15. The method which has emerged from this work replaces the entering cooling flow density term in equations (2) and (3) with the density of the heated air leaving the cylinders. This exit density is based on the flow static pressure and stagnation temperature.

An additional modification which is made is to utilize the density ratio "_" and absorb the sea level density constant into "a". The engine orifice equation now becomes )b.

w = a(Oexa p (4)

Equation (4) has been shown to work effectively up to

12,000 meters in altitude. Figure 14, taken from the

results of reference 14, shows a comparison of the extra-

polation capabilities of the exit density ratio versus a

density ratio based on the average of the ambient and exit

densities. The average density ratio parameter works for

low mass flows and low altitudes. However, the power law

breaks down with increasing mass flows and altitude. This

breakdown is caused by compressibility effects which be-

come dominant with higher flow velocities and altitudes.

Use of the ambient density ratio parameter solely would

result in even greater deviations. The exit density ratio _

power law, on the other hand, provides a valid extrapolation, unaffected by altitude, up to higher mass flows.

Therefore, in regard to the influence of heating and

altitude on ground test determined engine orifice character-

istics, there is a well proven relationship [equation (4)]

available to provide the necessary extrapolation. Altitude

orifice characteristics, derived from ground test data and

use of this relationship, should be considered as valid

data for flight test use.

Flight test measurements. The flight cooling air mass

flow measurement system is shown in Figure 15. The system

consists of an array of total and static pressure probes

mounted in each inlet and static pressure ports distributed

azimuthally in the inlet ducts. The inlets are axisymmetric

designs incorporated into the nose cowl to form the con-

ventional "bug eye" configuration. Previous studies of the

aerodynamic behavior of these inlets indicated that the

internal flow was well behaved and no adverse effects re-

suited from the propeller. The static and total pressure

distributions across the inlet duct showed some variation;

however, these variations were consistent with observed

angle of attack and propeller wake influences. No indica-

tion of flow seDaration or stall was observed in the data.

Using the least-squares technique, surfaces of the form

(5)

P = a z + a2v _ + a x 2 + a y2 + a x 3 + a6y3

1 3 4 5 were fitted to the total and static pressure data. Terms involving products of x and y were not usable in the poly- nomial because all probe location involved either x = 0 or y = 0 coordinates. Using the fitted polynomial for the static and total pressure distributions across the inlet duct, the corresponding cooling air mass flow rate was determined by numerical integration. Installation temper- ature and pressure data on both sides of the engine were also recorded.

Data were taken in altitude increments of 900 m from 1,800 m to 7,200 m. Airspeed and cowl flap settings were used to vary the mass flow. A sample of the data from three different altitudes is given in Figure 16. The baffle pressure drop value used here, and in subsequent graphs, is taken as the difference between upper plenum total and lower plenum static pressures. The reason for this will be discussed in a later section. The results are consistent with the behavior predicted by equation (4).

Taking the exit density ratio and placing it with the effective orifice area coefficient "a", one has b) (6) In (w) = b In (Ap) + In (a Oex .

Equation (6) represents a family of straight lines in logrithmic scaled coordinates. The last term on the right is the ordinate intercept which, as shown in Figure 16, varies with altitude. Using the corrected baffle pressure drop parameters, "_exAP '', the separated altitude curves collapse onto a single sea level curve as shown in Figure 17. The curve follows In (w) = b " In (OexA p) + In (a) Figure 18 presents a comparison between the result of Figure

17 and manufacturer's data applicable to the engines of the

test aircraft. The difference between the two curves lies

with the respective values of the effective orifice areas.

The "a" of the installed engine in the test aircraft is

significantly larger than that of the manufacturer's test

cell engine. The implication drawn was that approximately

55 percent of the cooling air entering the intakes was by-

passing the engine and leaking through the external engine

baffle system. Flow temperature measurements made directly

below the engine and at the cooling flow exit showed a

significant reduction, and suggested the mixing of heated

and unheated cooling air, lending support to the leakage

theory.

The external baffle system about the engine, shown in

Figure 19, is typical of current practice. A neoprene rubber

tape is used to provide the seal between the high and low

pressure sides of the engine. Figure 20 is a view of the

high pressure side of the cowled engine looking from front

to rear. The neoprene tape is seen laying against the inside

of the cowl as it is intended to do. Sealing is assumed to

occur when the tape is forced tighter against the cowl by the

ram pressure of the cooling flow. A simple and effective

method in theory, however, apparently not so in practice.

To test the leakage theory, the neoprene tape was re-

placed by a cover on top of the engine as shown in Figure 21.

This "do_ house" modification provided a positive lock seal

in this region, and removed the uncertainty that accompanies

passive systems such as the rubber tape. The flight tests

were reoeated and the results are given in Figure 22. The

"dog house" reduced the system orifice area, i.e., decreased

the leakage about the engine. With the "dog house", the same

engine baffle pressure drop is obtained with 38 percent less

cooling flow enterin_ through the inlets.

However, this improved curve still indicated sufficient

discrepancy from the manufacturer's curve to warrant additional

testing. Also, at this point, the question of the validity

of comparing the measurements between the flight installation

configuration and the ground test cell configuration came to

the forefront. To attack this question, a test technique

utilized in reference 4 was employed.

Ground test measurements. If an aircraft related investigation can be validly performed on the ground as well as in flight, ground testing should be strongly considered.

This is particularly the case when the investigation involves internal aerodynamics. Flight test investigations of cooling system aerodynamics are required if the concern is directed toward the external flow at the inlet and exit. Between these two points, however, flight serves only as a means of genera- ting an internal air flow, which could be accomplished just as well by a blower on the ground. Ground test investigations of

the internal aerodynamics of cooling installations offer the

advantages of unconstrained configurational variations and

flow measurements, and perhaps most importantly, the ad-

vantage of personal observation of the functioning of the

system. A moistened finger or cheek are very effective leak

detectors.

The ground test system is shown in Figure 23. The

system consists of a variable speed axial flow fan, mass

flow metering section, diffuser, and connecting ductwork

as required.

Tbe first configuration tested was the "dog house"

shown in Figure 21. The purpose of this test was to validate

the f!i_ht mass flow measurement system in the inlets. Com-

parisons between the inlet rakes and the duct metering

section showed a discrepancy of 17 percent. This correction

was applied retroactively to the flight test results. The

data in Figures 16, 17, 18, and 22 include this correction.

During this test, additional leakage was observed about the

valve, lifter covers and between the metal external baffle

and engine proper. These regions were sealed with duct tape

and silicone rubber, and the test was repeated. A reduction

in leakage of 8 percent was obtained by this additional seal-

Ing. With this configuration, leakage was still detectable,

primarily through the front engine baffle and out through

the gap between the prop spinner and nose cowl. The presence

of the nose cowl made it impossible to seal this region;

accordingly, it was removed and additional ductwork was

installed to reach the high pressure plenum of the engine.

This arrangement, shown in Figure 24, represents the max-

imum seal, no leakage, flight configuration case. This

configuration was important in that it served as the flight

configuration "ideal case" for comparison with the ground

test cell configuration. These two test configurations

are presented schematically in Figure 25. Testing of the

maximum seal flight configuration showed that the front

engine baffle was responsible for an additional 9 percent

leakage.

Referring to Figure 25, the ground test cell con-

figuration represents the "ideal case" from the standpoint

of flow behavior. The flow enters the high pressure side

of the engine in a uniform manner and in the same direction

as it will move through the cooling fin passages. The high

pressure plenum is large and the velocity is low. The

baffle pressure drop is measured as the difference between

the high pressure plenum static and ambient static. How-

ever, in the case of the flight configuration, the cooling

air enters at much higher velocities and in a direction

perpendicular to the fin passages. In many cases, it must

flow about obstacles directly in its path such as alternators,

oil coolers, prop governors, intake manifolds, etc. A

significant wake in the internal flow develops immediately,

due to the "bug eye" intake configuration. A true flow

static pressure is difficult to measure under these cir-

cumstances. Total pressures are relativel.y essy to acquire,

particularly with the use of Kie_ probes. The Doint of

concern here is whether the engine orifice characteristics

determined by ground test cell measurements are comparable

to those determined by flight test measurements. To in-

vestigate this question, a ground test cell configuration

was assembled and flow measurements performed.

The test cell configuration is shown in Figure 26.

Turning vanes were utilized in the bend and honeycomb in

the vertical duct to achieve an "ideal" uniform flow at the

engine. A total pressure survey made at the engine confirmed

a uniform flow. The engine baffle pressure drop was measured

in the same manner as standard test cell practice. The re-

sults from this test are given in Figure 27 alon_ with the

results from the maximum seal "ideal" flight configuration

test. The engine orifice characteristics as measured by

both test configur,qtions are identical if the baffle pressure

drop for the flight configuration is based on high pressure

plenum total, rather than on high pressure plenum static.

Use of the plenum static pressure gives an error of 8 per-

cent. The location of the total pressure probes was at the

rear engine baffle. Data presented in a later section show

this region to be the ].east affected by inlet configuration

and blockage. The key finding from this particular test

is that ground test cell determined engine orifice character-

istics are perfectly valid for flight test utilization if

the flight measurements are based on engine plenum total

pressure rather than static. Accordingly, all engine

orifice characteristic data presented herein utilize upper

plenum total and lower plenum static to form the baffle

pressure drop parameter. This includes Figures 16, 17, 18, and 22.

In order to achieve a valid comparison between the

flight and ground test results, a correction must be

introduced to convert the "cold" engine data to "hot"

engine data. In order to determine this correction, a

flight test was performed to measure the orifice character-

istics of both the "hot" and "cold" engine. The "hot" data

was taken with the engine operating at normal cruise power

settings. The engine was then shut down and allowed to

cool to ambient temperature, to obtain the "cold" data.

The "hot" and "cold" orifice characteristics are the same

if exit density ratio is used for the altitude correction.

However, if the ambient density ratio is used, the "hot"

curve shifts downward by I0 percent as shown in Figure 28.

Therefore, it appears that use of the exit density ratio as

a correlation factor, works well for both altitude correction, and "hot" and "cold" engine correction.

The results from the flight and ground tests are summa-

rized in Figure 29. It is evident from the figure that a

significant leakage problem exists with the use of the

rubber tape external baffle system. Considering the fact

that the test aircraft has very low time and, accordingly,

that the external baffles are in relatively good shade, one

can visualize the leakage problem which may exist with the

number of high time aircraft in service. Leakage produces

two adverse effects: increased cooling drag and reduced

cooling performance. The first of these is more obvious

since the internal flow drag is directly proDortional to

the mass flow of cooling air through the installation. The

second effect results from the reduction in the baffle

pressure drop which can be generated across the engine, due

to increased losses between the inlet and m!enum. The

operation of the inlets and the losses immediately down-

stream are functions of the total air mass flow, reFard-

less of whether it passes through or around the engine.

Reducing leakage will produce an increase in cooling flow

through the engine and, correspondingly, imDroved cooling.

Cooling Correlation Investigation

One of the many technological outgrowths of the radial

air-cooled engine development was the NACA Cooling Correlation

Analysis procedure. This procedure is well documented in

numerous NACA reports (see references 16 and 17). The pur-

pose of the method is to take ground based data of engine

cooling requirements and extrapolate these requirements to

operational altitudes. The NACA procedure required the de-

velopment of empirical relationships concerning the heat

generated by the engine, which could only be determined by

ground test. Consequently, use of the method for flight

investigations was always tied to the availability of ground

test results. This proved to be of little significance,

however, due to the close technological cooperation between

the government, airframe manufacturers and engine manufacturers

during World War II. This technological cooperation is today

much reduced in regard to the general aviation industry.

The competitive market situation in conjunction with product

liability concerns work against cooperation between airframe

and engine manufacturers in the solution of installation cool-

in_ problems. The airframe manufacturer, in many instances,

must work alone to solve cooling problems armed only with

ground test cell data, as represented by Figure 5.

A flight test program was performed to investigate

whether an installed engine cooling correlation procedure

could be developed to assist in the solution of cooling

problems. The objective was a cooling correlation relation

which involved parameters which could be easily measured

in flight. The NACA method was used as a basis (see

reference 16). The heat generated by combustion which

is transferred to the cylinders is given by

= WI (Tg - Th) (8) H ci "c

T!

The air charge flow "W c is directly relatable to the indicated engine power "I", and (8) can be rewritten as I TM _ (9) H = c 2 (Tg T h) .

The heat given up by the cylinders to the cooling air flow is k H = c3 w (T h - Ta). (I0) _ For flight testing, the cooling air mass flow "w" is imprac- tical to measure. The orifice characteristics of the engine are used here and the corrected baffle pressure drop parameter is substituted.

(ii) H = c4 (OexAP) n (T h - T a) For a constant cylinder head temperature, equations (9) and (II) must be equal, giving _ _ )n (12) (Th T a) / (Tg Th) clm/(Oex& p Equation (12) is essentially the NACA Cooling Correlation relation. The effective combustion gas temperature "T " g was determined by ground testing. Empirical relationships had to be established between it and other heat generating parameters such as fuel/air ratio, exhaust back-pressure, and ignition timing. As all other physical quantities in equation (12) are measurable in flight, it was decided to replace this term with the measured exhaust gas temperature.

For the PA-41P test aircraft, the supercharger turbine in- let temperature was used. Equation (12) now becomes (13) (T h - Ta)/(TEG T - T h) = T* = cIm/(cexAP) n The test program was flown as part of the orifice characteristics study. Altitude ranged from 1,800 m to 7,200 m. At each altitude, a test matrix of four power settings and three mixture settings was run. The mixture settings ranged from full-rich to peak EGT. Due to turbo- supercharging, the same four power settings were obtainable at all test altitudes. The cowl flaps were used to vary the cooling air flow at each test point.

Q The results are presented in Figures 30 and 31. In Figure 30, the relationships between engine baffle pressure

drop and the temperature ratio, T_, based on the average

of the cylinder head temperatures are plotted for a con-

stant indicated power. Each plot represents a range of

altitudes and cooling air mass flows. All four curves

have the same slope, but different ordinated intercepts,

dependent on the power setting. This is in agreement

with equation (13). Rewriting (13) in logrithmic form

with (OexAP) as the independent variable,

(14)

in(T*) = -n'In(oexAP) + In(cl m)

The intercept varies with the indicated engine power I.

The resulting cooling correlation relation is given in

Figure 31. A separate correlation was performed using

the hottest running cylinder temperature, (cylinder number 6), in

place of the averaged cylinder temperature. Results similar

to those of Figures 30 and 31 were obtained. Both cor-

relations are given below.

0 29

(15)

= _ ) = 0.1710"52/(OexAP) •

TA* (ThA- Ta)/(TEG T ThA

= ) = 0.1810.50/(aexAp)0.28 (16)

T6* (Th6 - Ta)/(TEG T - Th6

Both show excellent agreement with the behavior predicted

by equation (13).

The results of this investigation show that a cooling

correlation for a particular aircraft installation can be

developed in terms of quanitities which are easily measure-

able in flight. This correlation, once established, provides the basis for the solution of cooling problems,

and for relating cooling requirements to aircraft perfor-

mance. The airframe manufacturer is thus freed from

dependency on _round test cell data, for cooling in-

stallation flight test investigations.

Internal Flow Temperature Rise

As part of the coolin_ correlation analyses, correla-

tions were also developed for the temperature rise of the

cooling air at the rear of the upper plenum and at the

exit. The first of these locations is important in that

many installations use a portion of the airflow at this

point for auxiliary cooling, such as oil cooling, inter-

cooling, etc. The second location is of interest since it

leads to the exit density which is important for future

installation design analyses. The correlations were

developed using the basic formulation of equation (13).

Appropriate temperature terms were changed to represent

the heat transfer process. The results are given in

Figures 32 and 33. For the range of baffle pressure drop

(cooling air mass flow) developed by the PA-41P test air-

craft, the cooling flow temperature rise at the rear baffle

is

20°C < (Tup - Ta) < 30 ° , (17)

and the temperature rise across the engine is

(18)

70°C < (Tex - Ta) < 100°C.

The exit temperature rise values are consistent with

numbers reported by other sources for horizontally-opposed

engines. The upper plenum temperature values, however,

should be treated carefully. The heat transfer mechanism

here is commonly referred to as velocity cooling. The

amount of heat transferred to the cooling air is dependent

on the flow velocity and flow turbulence in the plenum,

and these, in turn, are dependent on installation and

engine configurations.

INLET INVESTIGATION

Background

The function of the inlet is to recover the available

dynamic pressure and deliver the cooling flow to the high

pressure plenum in a uniform manner. Ideally, this should

be accomplished with no internal or external flow separa-

tion. Inlets are classified as either two-dimensional,

as in the case of wing leading edge intakes, or three-

dimensional, as in the case of axisymmetric intakes. The

inlets used on general aviation reciprocating engine air-

craft are three-dimensional, and are of relatively complex

geometry in comparison to those used for turboprop and

turbojet powered aircraft. This is due primarily to the

configuration of the horizontally-opposed engine which

allows for an installation with minimal frontal area and

yet requires cooling air to pass vertically through the

engine. Conventional practice is to use two inlets, one

on eacb side of the propeller spinner. The design of the

inlet shape heretofore has been accomplished through a

combination of styling dictates and intuition. Very little

aerodynamic analysis and design is used. The reasons for

this are the absence of practical analytical methods and

the cost of aerodynamic testing of different candidate

shapes. As part of this program, a systematic study of

inlets and their effects on the cooling installation was

performed.

Inlet Aerodynamic Theory

The theoretical aerodynamic behavior of inlets will

be discussed in relation to three-dimensional axisymmetric

configurations. Figure 34 shows an inlet with the

pertinent design parameters identified. The purpose

of the inlet is to capture the required amount of flow

and recover a part of the dynamic head as an increase

in static pressure. The desired recovery of the in-

let itself is related to the velocity ratio (Vi/Vo),

i.e., the ratio of the inlet velocity to the free stream

velocity. A large pressure recovery corresponds to a

small velocity ratio and vice versa. The diffusion, through which the recovery occurs, takes place externally.

If sufficient internal length is available, then an

internal diffuser is also possible to increase the re-

covery.

The ability of an inlet to deliver the desired

pressure recovery and, in fact, its ability to function

properly over a range of operating conditions, is re-

lated to its cross-sectional shape as illustrated in

Figure 34. The axisymmetric inlets used in the inves-

tigation are from the KuchemannA-series, detailed in

reference 18. They consist of two distinct elliptical

segments which join at the nose of the lip contour.

The el]i_sesare in proportion according to the "A"

designation. Other inlet families are available for

different proportions. Three A-series inlets are shown

in Figure 35. The numerical part of the designation

4O

refers to the percentage ratio of the inlet area to

maximum external cross-sectional area. Low numbers

produce relatively thick lip contours with relatively

large radii of curvature and high numbers produce the

reverse. In Figure 36, the potential flow pressure

distributions about the lip contour are given for

different velocity ratios and angles of attack. The

distributions were calculated according to the method

of reference 19. Both the internal and external pres-

sure distributions exhibit characteristics similar to

those of airfoils. For many conditions, there is a

suction peak followed by an adverse pressure gradient,

which, when the boundary layer response is included, are

the necessary conditions for flow separation and stall.

The strength of the suction peak and severity of the

adverse gradient are dependent on the relative thickness

of the lip contour or, correspondingly, on the radii of

curvature employed. For larger velocity ratios, the

stagnation point is to the outside of the lip nose, the

suction peak is formed on the inside, due to the turning

angle in combination with the radii of curvature. For

small velocity ratios, the stagnation point lies to the

inside and the suction peak correspondingly forms to the

outside. In _eneral, therefore, inlets designed for

large velocity ratios and small external pressure recovery

are subject to the possibility of internal flow separa-

tion and stall. Inlets designed for small velocity

ratios and large external pressure recovery are like-

wise subject to the possibility of external flow sep-

aration.

The effect of the location of the stagnation point

on inlet pressure distributions suggests one fundamental

design tenet which should be followed for the reciproca-

tion engine cooling installations. With current technology,

the flow field about an arbitrary nose cowl shape is not

well defined. Therefore, the locations of the stagnation

points on the inlet lip contour are also ill-defined.

Accordingly, inlet lip contours should be well-rounded with

large radii of curvature to minimize inadvertent suction

peaks and following adverse pressure gradients.

Design of the Test Inlets

Five different inlet configurations were investigated

in this program. They are shown in Figures 37-41. They

consist of the original PA-41P inlet, three axisymmetric

configurations, and a conventional design representative of

current general aviation practice. The PA-41P inlet, here

designated as STD, is a swept configuration, with the purpose

of reducing propeller blockage by relieving the frontal area.

The propeller hub incorporates a 20 cm shaft extension to

provide the necessary length for this. The aerodynamic

effects of sweep for these types of inlets are unknown.

The three axisymmetric inlets were investigated because of

the existence of an experimental data base and analytical

design procedures. While this information applies to

true axisymmetric configurations, it provides a point of

reference for comparing the results of incorporating these

shapes into a cowl nose piece. The inlets are designated

according to their respective design velocity ratio

(0.3 or 0.6) and longitudinal location (forward or aft).

The fifth inlet is a conventional general aviation design

incorporating as much of the 0.3F inlet shape as possible.

The designation for this inlet is GAC.

Inlet Internal Effects

The STD, 0.3F, 0.6F, and 0.3A inlets were investigated

in one program phase. In a later phase, with some modifica-

tions to the external engine baffling, the 0.3F and GAC

inlets were tested.

The internal effects of the inlets are measurable in

terms of the pressure recovery obtained in the upper plenum

and the baffle pressure drop developed across the engine.

The baffle pressure drop varies directly with the cooling

air mass flow, as seen in Figure 5, and therefore is an

indication of the volume of cooling flow obtained. The

\ pressure recovery is important for two reasons. First, it represents one of the two internal losses in the system; the other being the baffle pressure drop. Poor pressure recovery, or large inlet losses, means less pressure head available for moving the cooling flow through the engine.

Second, poor pressure recovery results in higher cooling drag for the same cooling flow.

To properly evaluate the inlets, the nature of the flow immediately in front of the inlet needs to be defined. A total pressure survey rake was mounted close to the propel- ler disk as shown in Figure 37(a). The results from this survey shown are given i_ Figure 42 in terms of the pressure coefficient referenced to free stream static. Superimposed is a planform view of the nacelle. The effect of the propeller is seen in terms of the distribution of total pressure available at the inlet. This varies from 0.85q_ at the spinner to approximately l.lq= at the outer edge for cruise, and to 1.2q= for climb. The reductions in total Dressure near the spinner are attributed to losses associated with the propeller shanks operating in a stalled condition, i.e., the shank angle of attack is too high for the operating condition. On the average, a loss of 0.1q= occurs for the inlet flow in cruise.

Figures 43-46 give results for the upper plenum total pressure surveys for four of the test inlets. The survey

points were at the end of the inlet duct and across each

row of cylinders (see Figures 8, 9, and I0). The presenta-

tion shown is a top view of the engine with the inlets in

the upper left and upper right corners of the graph. The

respective cylinders are denoted by the numbers 1-6 outside

the graph. The propeller governor is located in the lift

inlet in front of cylinder number 2.

The 0.3F inlet in Figure 44 gives the highest pressure

recovery in the plenum. The inlet ducts are well behaved

as indicated by showing a total pressure distribution similar

to the propeller rake. The 0.6F, 0.3A, and STD inlets re-

sulted in lower values of upper plenum total pressures.

The pressure recovery of the plenum is obtained through a

combination of external diffusion by the inlet, internal

diffusion by incorporating an expanding area diffuser in the

inlet duct, and the diffusion associated with the flow ex-

panding into the plenum volume fromthe inlets.

The STD inlet in Figure 43 shows a loss in pressure

recovery in its left inlet, indicating the presence of an

internal stall. This may be due to either the blockage by

the propeller governor located in the left inlet, or due

to the left inlet operating at a higher angle of attack

than the right inlet and consequently exceeding the design's

stall angle. The angle of attack asymmetry is due to the

swirl component of the propeller flow. Also evident in

the figure is the nonuniformity of the flow entering the

plenum as indicated by the variation in total pressure.

This again is the result of insufficient diffusion by the

inlet.

The 0.3F inlet in Figure 44 results in good pressure

recovery in the plenum. The flow is more uniform, in-

dicating that part of the recovery has been accomplished

externally by the inlet.

The 0.3A inlet in Figure 45 produces 0.2q_ less pres-

sure recovery than the 0.3F. This loss is believed to re-

suit primarily from the poor interface between the inlet

duct and the plenum entrance. The inlet duct is partially

obstructed by the front cylinders. The extent to which

external diffusion was accomplished is unknown. As the

inlet moves aft, its geometry, which controls its aero-

dynamics, is increasingly compromised by the nose cowl

geometry.

The 0.6F inlet in Figure 46 shows similar poor recovery.

While the 0.3F inlet was designed for external diffusion,

the 0.6F was designed for internal diffusion using an

appropriately configured internal duct. For both climb and

cruise conditions, the inlet indicates an internal stall by

the loss in pressure in the region adjacent to the spinner.

This condition is more severe for climb than for cruise.

The inlet is operating at a higher velocity ratio in climb

due to the use of cowl flaps to pump additional flow through

the system. Referring to the inlet velocity ratio contours

in Figure 36, this increases the internal suction peak and

following adverse gradient, thereby increasing the tendency

towards internal stall.

The GAC inlet was tested at a later time after the "dog

house" modification had been made to the external engine

baffle. Cooling air mass flow measurements had shown the

external baffle to have considerable leakage. The four in-

lets just discussed were operating at significantly higher

velocity ratios than anticipated. This aggravated internal

aerodynamic problems. The 0.3F inlet was also operated with

the modified baffle and consequently it is used as a reference

for comparison purposes. The results are given in Figures

47 and 48. The 0.3F inlet shows a slight improvement over

the earlier test. This is due to reducing the flow losses

associated with the inlet duct interface with the plenum.

These losses were reduced as a result of reducing the cool-

ing air flow by elminating the baffle leakage. The GAC inlet

in Figure 48 shows about the same performance as the STD in-

let. The absence of the survey rakes at the end of the

inlet duct for these tests increases the difficulty of

interpreting the results. The indication is that, like the

other low performance inlets, little or no external diffusion

occurred and significant flow losses were created at the

plenum entrance. Additional information concerning the

contour pressure distribution are given in a later

section.

The effects of the inlets on cooling installation

performance over the complete operating range of aircraft

are given in Figures 49-52. The parameters of importance

are the plenum pressure recovery and the engine baffle

pressure drop, which is directly related to the cooling

air mass flow rate. These parameters are presented in

Figure 49 for climb at a constant I00 kts equivalent air-

speed. Only the three inlets shown were subjected to a

climb test. The 0.3A inlet did not allow adequate cooling

for this flight condition. The indicated altitude dependen-

cies are due to propeller effects. While maintaining the

same engine power in the climb, the propeller survey rake

indicated a reduction in slipstream total pressure with

altitude. This reduction in slipstream total pressure

caused a reduction in plenum pressure recovery and also

reduced pumping effectiveness of the cowl flaps.

The 0.3F and STD inlets result in the same baffle

pressure drop, i.e., the same cooling air mass flow. The

0.3F inlet, however, accomplished this at a higher pressure

recovery which translates into lower internal cooling drag.

There is more energy in the flow at the exit than for the

STD inlet. The decrement in baffle pressure drop pro-

duced by the 0.6F inlet is due totally to its loss in

pressure recovery.

Baffle pressure drop and plenum pressure recovery for

cruise conditions are shown in Figures 50 and 51. These

parameters are shown to be independent of altitude and

only slightly dependent on airspeed. This dependency

appears to be due to changes in the external nacelle

pressure at the exits which accompanies the angle of attack

variation with airspeed. In Figure 50, the 0.3F and STD

inlets generate approximately the same baffle pressure drop,

while the 0.3A and 0.6F inlets result in somewhat lower

capability. The marked distinction is in the pressure

recovery, wbich exerts an important influence on cooling

dra_ as well as cooling air mass flow. The 0.3F and GAC

inlet results, in Figure 51, were tested after the modifica-

tion to the engine baffles to reduce leakage. An increase

of 0.2q_ in baffle pressure drop is seen for the 0.3F in-

let as a result of this change. The GAC inlet appears to

function at the level of the 0.3A and 0.6F inlets.

The effect of propeller operation on inlet performance

is shown in Figure 52. Improvements in pressure recovery

and baffle pressure drop are obtained for the 0.3F, 0.6F,

and 0.3A inlets, while the GAC inlet shows no change, and

the STD shows a reduction with propeller running. The

behavior of the STD and GAC inlets is not understood in

this regard. It is believed that the answer lies with

the interaction between the swirl component of the slip-

stream and the particular inlet geometry.

External Inlet Effects

Flow visualization. The external flow about the in- lets was investigated through the use of in-flight tuft photographs. Pressure distribution data were also ob- tained for the 0.3F, 0.6F, 0.3A, and GAC inlets. Initially, a wide range of flight conditions, coolin_ air flows, and propeller operating conditions were run. After observing that the external flow behaved in a systematic manner, the test conditions were reduced to representative climb and cruise conditions, each with propeller running and propeller stopped.

The in-flight tuft studies are presented in Figures 53-57. The STD inlet in Figure 53 is shown with the propeller stopped as well as running. As also indicated in the other figures, the nacelle is operating at a positive angle of attack for the cruise condition. The nose cowl stagnation point appears to be immediately below the propeller spinner. A strong upward flow is indicated aft of the spinner and a strong outward flow is shown below the inlets. The flow into the inlet has an upward component at the spinner. The inlet flow is

separated on the lower intake contour, as a result of the

small radii of curvature used there. As will be seen

with the pressure distribution results, with propeller

stopped, the stagnation point on the upper lip tends to

move toward the inside. This produces a suction peak, following adverse gradient and subsequent local separation.

This is evident on the outboard inlet. The inboard inlet

is already separated with propeller running. Stopping

the propeller increases the separated area.

The external flow about the 0.3F and 0.6F inlets in

Figures 54 and 55 is well ordered. The flow appears to

be primarily in the longitudinal direction with no obvious

lateral or azimuthal components, and no flow separation

The 0.3A and GAC inlet configurations in Figures 56

and 57 show behavior similar to the STD configuration.

The stagnation point below the spinner causes upward flow

on the nose cowl immediately aft of the spinner, and a

flow below the inlet. The intake area appears to be

stalled for both inlets. The external flow about the

inlets is unseparated and orderly.

Inlet pressure distributions. Figure 58 shows the locations where inlet pressure distribution data were taken. The results for the cruise condition for the 0.3F, 0.6F, 0.3A, and GAC inlets are presented in Figures 59-62.

Pressure data for the STD inlet were not acquired. For

the cruise condition, data were taken for the propeller

running under load, and for propeller stopped and feathered.

The pressure distributions for the 0.3F inlet are

given in Figure 59. Referring to Figure 36, the pressure

distributions are consistent with the axisymmetric inlet

design. The difference between the upper and lower sur-

faces indicates that the inlets are operating at a positive

angle of attack. The difference between the inboard and

outboard inlets is believed to be due to blockage of the

inboard intake duct by the propeller governor which causes

this side to operate at a lower velocity ratio than the

outboard. With the propeller stopped, the stagnation

points on the inboard inlet move to the inside which is

consistent with the reduction in cooling air flow that

accompanies this. The outboard inlet shows little response

to stopping the propeller, indicating no appreciable move-

ment of the stagnation point. The change in velocity here

appears to be small. The side pressure distributions show

asymmetry when the propeller is stopped. This behavior

is evident on all of the inlets. It is believed that this

is due, in part, to side-slip angles which resulted from

trimming the aircraft after shutting down the right engine.

The 0.6F inlet in Figure 60 shows similar behavior.

While this inlet was designed to operate at a higher

velocity ratio than the 0.3F inlet, the pressure distributions

do not reflect this. They are more peaky than the 0.3F, indicating a lower operating velocity ratio. However,

the internal total pressure surveys showed this inlet to

be stalled in both intake ducts, so that the results of

Figure 35 no longer apply. Accordingly, it is impossible

to interpret the pressure distributions beyond this point.

The 0.3A inlet in Figure 61 also demonstrates similar

behavior. As with the previous inlets, the external pres-

sure distribution is consistent with the axisymmetric

geometry, and there is no indication of separation of

the external flow.

For the GAC inlet, a more extensive internal pressure

distribution was obtained. The results are given in

Figure 62. The internal pressure data indicate attached

flow and an initial pressure recovery of 0.8q . The motion

of the stagnation point on the upper lip contour is noted

with stopping the propeller. Unlike the axisymmetric

configurations, both the inboard and the outboard inlets

exhibit similar behavior, with some difference showing

for the lower contour. The side pressure distributions

differ from the axisymmetric models for the stopped

propeller case. The stagnation point moves towards the

inside for both inlets. The outboard lateral distribution

indicates a local stall. Again, because of the complex

geometry involved, it is impossible to explain this

behavior further without an appropriate analytical model

or additional experimental data.

EXIT INVESTIGATION

Background

The exits of the cooling installation act as the

system throttle. Since the flow is subsonic throughout,

the quantity of cooling flow will adjust itself so that

the pressure at the exit is equal to the local external

flow pressure. The relationship between cooling flow

volume rate and exit pressure is governed primarily by

the exit duct area. Too small an exit area will throttle

the flow and lower the volume rate. Too large an exit

area may result in mixing problems between the cooling

flow and external flow, which can have an adverse effect

downstream. An additional function, which must be in-

corporated into the exit configuration, is a cooling flow

pumping mechanism for the climb flight condition, where

the flow volume required for cooling at high power

settings cannot be generated by the low flight velocity.

The mechanism most often utilized is a cow] flap, which

is essentially a spoiler, sized to produce a local low

pressure wake immediately downstream.

Exit configurations and locations are also well

standardized. The predominant location is on the lower

surface of the cowl, well back from the nose. The local

external pressure is generally close to free-stream static.

Dependin_ on landing gear placement and other related

considerations, the exit may be split into two ducts, one

on either side of the cowl, or a single duct may be used

on the bottom. Periodically, the exits have been located

on the upper surface of the cowl, particularly for twin-

engine configurations where there is a low pressure region

available for pumping. A negative aspect of the upper

surface location is the depositing of oil and grime, which

is picked up by the cooling flow, on parts of the airframe

which may come in contact with the passengers and/or crew.

Exit Test Configurations

The original PA-41P exit configuration is shown in

Figure 63. This is a split system located on the lower

cowl beneath the leading edge of the wing. The cowl

flap is shown fully deployed. The existing installation

left little room for configurative variations. The central

area of the lower cowl was occupied by the landing gear

and oil cooler. This, in combination with the requirement

to maintain fire wall integrity, dictated that existing

location be used. Three exit system parameters could be

varied: exit area, cowl flap deflection, and cowl flap

aspect ratio. The relationship between cowl flap de-

flection and exit area depends on the location of the

flap hinge in relation to the exit duct. Figure 64

shows two examples of this arrangement, one where cowl

flap deflection increases the exit area, and the other

where exit area remains constant. The restrictions im-

posed by the PA-41P installation dictated that the first

arrangement be used.

The exit design parameters which were investigated

consisted of exit area and cowl flap aspect ratio. The

test cowl flaps were relocated so that the original PA-

41P exit area was increased by fifty percent. Restrictor

fairings were then placed in the exit duct which reduced

the area to its original value and then to a fifty percent

decrease below this. The exit area variation is shown in

Figure 65. Three cowl flaps, with aspect ratios of 1.5,

0.75, and 0.55, were tested with each of the exit areas

to produce a 3 x 3 configuration matrix. The cowl flaps

were installed by relocating the hinge line so that all

three produced the same exit areas at the same settings.

This, however, resulted in different cowl flap deflection

angles with the short, high aspect ratio flap having the

largest deflection angle. The three cowl flaps are

shown in Figure 66. The range in deflection angles

between the lon_ and short flaps is shown in Figure 67.

All pertinent geometric data are presented in Table II.

Exit Area Test Results

The results from the exit area investigation are

given in Figure 68. Application of Bernoulli's equation

shows that the lower plenum pressure is inversely pro-

portional to the exit area. Since the combined pressure

drop from inlet to exit is constant for any specific

flight condition, a reduction in lower plenum pressure

means an increase in baffle pressure drop can be accom-

modated, and accordingly, an increase in cooling flow.

The variation in lower plenum pressure is seen in this

figure. The change in lower plenum pressure is not con-

verted completely into baffle pressure drop, however.

With the increased cooling flow the upper plenum pressure

also decreases, because the inlet pressure recovery is

reduced with the higher velocity ratio and the internal

inlet losses increase with increased flow velocity. It

is clear from Figure 68, however, that baffle pressure drop,

andcorrespondingl_ cooling flow can be increased by in-

creasin_ exit area. The implication is that poor inlet

design of leaky baffles can be compensated by subsequently

increasing the exit area to achieve the required cooling.

However, internal cooling drag will be increased through

increase in momentum defect, and external drag mav be in-

creased through mixing with the external flow and associated

deleterious effects downstream.

Cowl Flap Test Results

The results from the cowl flap study are presented in

Figure 69. Due to cooling requirements, this study was

performed at a low speed cruise condition rather than climb.

The prime difference here is the absence of the additional

slipstream velocity which influences the effectiveness of

the cowl flap. The results show a decrease in lower plenum

pressure and an increase in baffle pressure drop with cowl

flap deflection. The spread in the curves as the cowl flaps

are closed is opposite to what would be exDectedo In the

closed position, the exit area should be the same for all

three flaps. During the test, it was observed that as

the length of the flap increased, the hinge moments on the

flap, in the closed position, also increased. The posi-

tioning linkage of the cowl flaps allowed some defection

under load. This increased the closed position exit areas

somewhat above the values in Table II. It is believed,

accordingly, that the spread in the results at the closed

position is due to the medium and long cowl flaps deflecting

towards the open position due to aerodynamic loads. The

cowl flap airloads vanished as the flaps were deflected

towards the open position. The Table II values are

correct for the no airload open condition.

In the open position, no difference is seen between

the three cowl flaps in Figure 69. Considering the wide

deflection angle range, as listed in Table II, it appears

that exit area, not deflection angle, is the controlling

mechanism here. The deflection of the cowl flap is supposed

to generate a low pressure region immediately downstream

which acts as a pump for the cooling flow. This effect is

not apparent for the configurations tested.

Exit Location Investigation

The interest in the use of upper surface cooling air

exits is driven by two basic ideas. First, for engines with

updraft cooling, i.e., where the cooling flow enters the

bottom and exits the top, upper surface exits are more

expedient, rather than requiring the flow to return again

to the lower surface for exhausting. Second, particularly

for twin engine aircraft, there are obvious regions of low

pressure which seem to offer the advantage of additional

pumping to increase the baffle pressure drop across

the engine. This ignores the fact, however, that the

flow in a low pressure region on an aerodynamic body, must

ultimately negotiate an adverse pressure gradient to reach

free stream static pressure. Due to boundary layer

effects, this is difficult to achieve without flow

separation in the most favorable of circumstances.

The introduction of a low momentum secondary flow in-

to a low pressure region can well result in severe

effects downstream in terms of flow seDaration and

corresponding increases in drag.

Nacelle pressure distribution measurements. Prior to locatin_ the upper surface exits, nacelle pressure distribution measurements were taken. Figure 70 shows the Doints where static pressure belts were positioned for the longitudinal pressure distribution measurements. The sym- bols in the figure relate the indicated position to the pressure data given in Figures 71 and 72. The results pre- sented in Figure 71 and 72 show the nacelle external static pressures at the exits, on the lower surface, and at the longitudinal point of lowest pressure on the upper surface.

The low pressure point coincided with the suction peak of the wing section near the leading edge. In Figure 71, the results are for cruising flight. A noticeable difference exists be- tween the nacelle pressures on the inboard side and the outboard side. Due to the presence of the fuselage, the flow in this region generally has higher velocities than comparable points

outboard of the nacelle. Also, on the starboard side of the

aircraft, the swirl of the propeller slipstream increases the

angle of attack of the inboard wing section, and decreases

the angle of attack of the outboard wing section. This also

contributes to higher velocities and correspondingly lower

pressures on the inboard section. Figure 72 presents the

results for climb power at different airspeeds. Also in-

cluded for comparison purposes are the same data for pro-

peller stopped. At the best climb speed of approximately

I00 kts, the propeller slipstream amplifies the pressure by

about 0.4 q . In both cases, the lowest pressure is found

adjacent to the upper wing surface. The pressure increases

towards free stream as one moves away from the wing as in-

dicated by the results for the top of the nacelle. In

cruise, there is a potential doubling of the suction nres-

sure at the exits bv locating them on the upper surface.

Exit location results. The installation of the upper surface exits is shown in Figure 73. The exit area was set at the largest of the three investigated in the lower surface exit investigation (150% of original). Louvers were used at the exits to align the exit flow with the external flow. The final test configuration is shown in Figure 74. During the test program, the lower exits were closed off and all cooling

flow passed through the upper exits. The results are given

in Figure 75 for both cruise and climb conditions. Three

sets of data are presented: upper surface exits with louvers

on, louvers off, and conventional lower surface exits. The

upper surface exits increase the baffle pressure drop across

the engine by 0.1q . The louvers contribute some flow resis-

tance which affects the obtainable baffle pressure drop. In

the climb condition, the upper and lower exit systems are

essentially equal. However, here the lower exits have been

modified by deployment of the cowl flaps. The upper surface

exits are shown to be superior in cruise and comparable in

climb to the conventional lower surface exits. Intuitively,

one would expect the upper surface system to offer less drag

in climb than the lower surface system with cowl flaps open.

A drag study was performed comparing the upper and lower

surface exits in both climb and cruise configurations. Rate

of climb was used as the measure for the climb test. The

results were inconclusive. No judgement could be rendered

concerning the relative drag of the two systems. The gen-

eralized speed/power method was used for the cruise config-

uration. The results are given in Figure 76 in terms of

generalized power versus generalized velocity. The upper

surface exits result in a definite drag increase for the

aircraft. In terms of generalized velocity, there is

approximately a 6 kt decrease in speed for the same power

setting. During the test program a mild buffet was felt in

the horizontal tail due to the flow from the upper surface

exits. With the louvers removed, the buffet increased

noticeably, indicating the presence of a well developed wake.

Tuft studies did not indicate any significant change in flow

patterns. Other than its manifestation through tail buffet,

the actual formation and nature of this wake was not deter-

mined. However, the indicated presence of the wake lends

support to the drag measurement results. For the config-

uration tested, the cooling installation benefits of the

upper surface exits were negated by an associated drag in-

crease. A similar configuration was tested in a full scale

wind tunnel (reference II). The drag results were in agree-

ment with the results reported here. Reference 1 also in-

dicates that locating the exits in a low pressure region does

not lead to the best configuration when the resulting drag

is included in the evaluation. In summary, the reported

results to date, do not support the apparent benefits of

cooling air exits in low pressure regions of the aircraft°

CONCLUSIONS

I. With present techniques, reliable measurements of

cooling drag are difficult to obtain through flight

test. Such investigations should be performed through

wind tunnel test whenever practical.

2. Much of the radial engine technology concerning engine

orifice characteristics and altitude correlation is

directly applicable to horizontally-opposed engines.

3. The differences between ground test cell configurations

and flight installation configurations, in regard to

internal aerodynamic measurements, are accounted for by

using total pressure measurements rather than static

pressure measurements. The measurements should be of

the oncoming flow from the inlets rather than that of

the component passing through the engine.

4. Current design practices of using a rubber tape lap seal

for the external engine baffle results in significant

leakage, cooling problems, and increased cooling drag.

5. A simple ground test blower system has been shown to be

an important tool for the development of aircraft

cooling installations.

6. A flight test technique for the determination of the

installed engine cooling requirements are determined in

terms of easily measurable parameters, thus freeing the airframe manufacturer from the restrictions imposed by ground test cell data in this regard.

.

The aerodynamic behavior of the inlets are a major factor in the effectiveness of the cooling installation. There is an obvious need for basic inlet design guidance.

, The design parameter of exit area has been shown to agree with theory in regard to its effect on the cooling installation. The locating of the exits in a low pres- sure region should not be attempted without a thorough study of the consequences, k

TABLE I. - COOLING DRAG FLIGHT TEST RESULTS

Configuration C D @ CL = 0

0.0226

No cooling flow

0.0234 (+4%)

Augmentor 25% open

0.0226 (+0%)

Augmentor 50% open

0.0242 (+7%)

Augmentor 100% open

TABLE II. - COWLFLAP TEST CONFIGURATIONDATA

Cowl Flap Deflection (deg.)

Cowl Exit Area

(cm2 )

Flap Position

Short Medium

Long

1-closed 214

2 268 13 5 3

3 300 22 I0

4 326 31 14

18 13

366 42

5-open

REFERENCES I. Hammen, T. F., Jr.; and Rowley, W. H.: Factors Pertaining to Installation of Inverted, In-Line Aircooled Aircraft Engines. SAE Journal (Transactions), Vol. 54, No. 3, March 1946.

2. Ellerbrock, H. H., Jr.; and Wilson, H. A., Jr.: Cowling and Cooling Tests of a Fleetwings Model 33 Airplane in Flight. NACA WR L-632, 1945.

(Formerly NACA MR May 1944).

3. Conway, R. N.; and Emmons, M. A., Jr.: An Investigation of the Ranger V-770-8 Engine Installation for the Edo XOSE-I Airplane. 1-Cooling. NACA WR L-561, 1945.

(Formerly NACA MR L5112).

4. Nichols, M. R.; and Dennard, J. S.: An Investigation of the Ranger V-770-8 Engine Installation for the Edo XOSE-I Airplane. ll-Aerodynamics. NACA WR L-562, 1945.

(Formerly NACA MR L5112b).

5. Nichols, M. R.; and Keith, A. L., Jr.: An Investigation of the Cowling of the Bell XP-77 Airplane in the Propeller-Research Tunnel. NACA MR November 1943.

6. Nielsen, J. N.; and Schumacher, L. E.: Analysis of the High-Altitude Cooling of the Ranger SGV-770 D-4 Engine in the Bell XP-77 Airplane. NACA CMR October 1943.

7. Monts, F.: The Development of Reciprocation Engine

Installation Data for General Aviation Aircraft. SAE

Paper 730325, April 1973.

8. Miley, S. J.; Cross E. J., Jr.; Owens, J. K.; and

Lawrence, D. L.: An Investigation of the Aerodynamics

and Cooling of a Horizontally-Opposed Engine Installation.

SAE Quarterly Transactions, Vol. 86, September 1978.

9. Miley, S. J.; Cross, E. J., Jr.; Owens, J. K.; and

Lawrence, D. L.: Aerodynamics of Horizontally-Opposed

Aircraft Engine Installations. AIAA Paper 77-1249, August 1977.

I0. Miley, S. J.; Cross, E. J.,Jr.; Ghomi, N. A.; and

Bridges, P. D.: Determination of Cooling Air Mass Flow

for a Horizontally-Opposed Aircraft Engine Installation.

SAE Quarterly Transactions, Vol. 88, August 1980.

Ii. Corsiglia, V. R.; Katz, J.; and Kroeger, R. A.: Full-

Scale Wind Tunnel Study of Nacelle Shape on Cooling Drag.

AIAA Paper 79-1820, August 1979.

12. Rubert, K. F.; andKnopf, G. S.: A Method for the Design

of Cooling Systems for Aircraft Power-Plant Installations.

NACA WRL-491, 1942.

13. Biermann, A. E.: The Design of Metal Fins for Air-Cooled

Engines. SAE Journal (Transactions), Vol. 41, No. 3, 1937.

14. Goldstein, A. W°; and Ellerbrock, H. H., Jr.: Compressi-

bility and Heating Effects on Pressure Loss and Cooling

of a Baffled Cylinder Barrel. NACA Report No. 783, 1944.

15. Neustein, J.; and Schafer, L. J., Jr.: Comparison of

Several Methods of Predicting the Pressure Loss of

Altitude Across a Baffled Aircraft-Engine Cylinder.

NACA Report No. 858, 1946.

16. Pinkel, B.; and Ellerbrock, H° H., Jr.: Correlation of

Cooling Data from an Air-Cooled Cylinder and Several

Multi-cylinder Engines. NACA Report No. 683, 1940.

17. Manganiello, E. J.: Valerino, M. F.; and Bell, B. B.:

High-Altitude Flight Cooling Investigation of a Radial

Air-Cooled Engine. NACA Report No. 873, 1947.

18. Kuchemann, D.; and Weber, J.: Aerodynamics of Propulsion.

McGraw-Hill Book Co., Inc., New York, 1953.

19. Stockman, N. O.; and Button S. L.: Computer Programs

for Calculation Potential Flow in Propulsion System

Inlets. NASA TM X-68278, 1973.

I J (a) radial (b) in-line Figure I. - Aircraft engine cooling installations.

7O

"--. ', ©...o..o...O..O...q

(c) vee I (d) horizontally-opposed Figure I. - Concluded.

C)

Q; Q; f_ bO r..4 C; .r.4 Q; 4--} q-4 (.; pC_ I -,.n'- o") o Q; pc_ c,,l Q; b_ .,.4 4J _4 CJ 4J !

!

b_ Figure 4. - PA-41P test aircraft cooling installation.

co I- _7 _I:I,I o I,III: o -r_ I r,-_ o o ILl hl i-i l.iJ t_l\ tl:)_ Ill n" _: i:I. c'xl\ e_l\ o

o _

,,=

i,_ -I:I.

i i\ \i_

,.o 0 I-- n_

oo

' Z oas / 8_l - MO-I:I _IIV 9NI-IO00 \ 0 _ 4-1 0 Io 0 i_ ._i 0 0 0 0 0 0

5 ,,7 o o o

0 _ o bO I

._o.N

,H

"'° _

.0 i

®

® ---- ®-___ ____

---- ___ ® ----

®

Figure 6. - Cooling installation model schematic.

Figure 7. - Engine orifice characteristics and cooling requirements determination test set-up, KIEL TUBES

I FI I

_OTOT_ilitl IIIIIIII

I I D D

!I I I I

if I,,lllllllJ

i]Jilll ENGINE

ii

TOP VI EW lllt;t_

lllli_JillZ

0 I I

I I n

ii II illllll_

_,lJllll®

llilll_

lllltIIIliil

I I n n

D I I I

,,II

JJ it II IIIIIi

®

iiiii ill-ill

iilll,, ® I TEMPERATURE THERMOCOUPLE PICCOLO TUBE Figure 8. - Location of Kiel total pressure probes and temperature probes in the high pressure plenum.

4--) ,--M "H 4-J .r-I O t_ O (D g} (9 _L 4-J O 4.J (D I ,r--I bO .,-I 4-1 4-3 4J "l:J UI 12u r-4 4-1 4-1 4-1 01--t 4-1 4-1 m _-4 12U I o (1) o_

,,.,,IILIIIIL