Aircraft Intake Aerodynamics Revised 4/22/2014 13:37:00 Divas Gupta 1 Arjun Agrawal 2 Prabhat Kumar 3 1 Department of Mechanical Engineering, Indian Institute of Technology, New Delhi Delhi,, INDIA 110016 e-mail: [email protected]2 Department of Mechanical Engineering, Indian Institute of Technology, New Delhi Delhi,, INDIA 110016 e-mail: [email protected]3 Department of Mechanical Engineering, Indian Institute of Technology, New Delhi Delhi,, INDIA 110016 e-mail: [email protected]1. INTRODUCTION In power systems design of an aircraft proper selection and design of a suitable intake system is very crucial for smooth running on the engine at all times (take-off, cruise, landing). For proper combustion to happen in an engine the flow in the combustion chamber should be almost stagnate and this calls for inclusion of a diffuser in front of the combustion chamber. The main aim for an intake design is: 1. Make sure that air is available to engine at all times without any suffocation. 2. To minimize the losses in total pressure along the path from free stream to the combustion chamber as this will tend to increase the amount of charge entering into the engine and will in return produce equivalent amount of power. 3. Also minimize the drag acting on the intake system due to external flow around the engine. Most of the time in this paper, we will be talking about the flow losses (total pressure loss) due to an intake system which will be quantified by: Pressure recovery factor, 1 i P q We will look into the cases of subsonic, transonic and supersonic intake flows. 2. SUBSONIC INTAKES 2.1 Types of Intakes Two Categories based on location of engine on the air craft: i. “Podded” installations Boeing 707 Engine Source: airliners.net ii. “Integrated” installations British Aerospace Harrier GR7 Source: flugzeuginfo.net Three Categories based on Method of intake in subsonic flows: i. Direct or fully ducted intakes ii. Plenum chamber installations iii. Propeller-turbine engines
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
2.2 Approximate theory of frictional losses p+1/2dp Following analysis done by Seddon (1952).
Assumptions: A+dA
- No flow separation, only frictional loss A,v,p dl v+dv - Inclination of boundary of streamtube to its axis is small. p+dp
- g is the local perimeter which may be or may not be the
complete perimeter of that station.
Fig 2.1 : streamtube with boundary viscosity.
F is therefore equal to q Cf g dl, where q=1/2 V2, and Cf is the local friction coefficient.
Applying the momentum theorem to the streamtube element of initial area A gives:
Av(v + dv – v) = pA + (p + ½ dp)dA – (p + dp)(A +dA) – F
Whence
Av dv = - Adp – F
Or
d(p + ½ v2) = -F/A
dP = - F/A,
where,
Total Pressure, P = p + ½ v2
P =
2 2
1 1
/ fg
F A qC dlA
Non dimensionally and using continuity equation (Av=1A1v1)
2 2
1 1 2
1 11 1
( )f fP q g A g
C dl C dlq q A A A
(2.2)
We have an approximate expression for loss of total pressure along a streamtube bounded wholly or partly by a
solid surface.
Wetted area, S : for an intake on the side of the fuselage, S is taken to be the surface area between entry to the engine
to the foremost point of the fuselage nose which comes in contact with the inlet flow to the engine. This definition is
somewhat imprecise since actual streamline patterns ahead of the intake are not usually known precisely.
Consider now the flow into an intake:
fuselage nose
A,v Ac,vc
Af,vf
0 lc
lf
v
l = 0 to l = lc : Approach
l = lc to l = lf : Duct v
0
Fig 2.2. Representation of internal flow with external wetted surface
Approach Duct
Pre-entry
Retardation
Taking q1 = qc , dynamic pressure at entry, and considering incompressible flow since subsonic, we can write:
2
0
( )
flc
f
c
P A gC dl
q A A
(2.3)
Approach Loss :
2
0
( )lc
a cf
c
P A gC dl
q A A
(2.4)
Duct Loss: 2( )
fld c
f
clc
P A gC dl
q A A
(2.5)
Approximation for approach loss: 3.( )a
a cF
c c
P A SC k
q A A
Where, CFa – overall friction coeeficient of approach
S = gdl
k – empirical factor taken to incorporate the difference from v of the assumed contant approach velocity.
We take, corrected position ratio, J = kS/Ac & hence,
3
.a
a cF
c
P AJC
q A
Approximation for duct loss: 2
. ( )
fld c
Fd Fd
clc
P A gC dl IC
q A A
Where
Duct Integral, 2( )
flc
lc
A gI dl
A A
& CFd – effective overall friction coefficient whose value can be defined for a conical diffuser from fig 2.3 and
following correlations:
f
2( )tan
2 ( )
c f f c
c f c
a g A g A
g g l l
& e
4R
cff
c
Ac V
g v
Reff is used to calculate Cf for flat plate used in the graph.
Fig 2.3 – Effective friction coefficient in duct deduced from
experiments of “Squire”. 1
a
Total Loss: 3
Fd Fa
c
PIC JC
q
(2.6)
=Ac/A which is inverse flow ratio
( 1)
ps pm
f MC C
J
2.3 Examination of variation
We compare the results for Wing leading-edge intakes for various angles of sweep. Sweepback angle,
ICFd JCFa k
0 0.14 .010 .80
30 0.11 .021 .65
40 .08 .033 .72
52 .05 .054 .80
variation agrees with the experimental results for >1 and for some values below =1 except for large
sweep back angles. This is due to the inclusion of additional loss of flow separation. This additional loss
reaches a maximum in ground running conditions, =0, when the engine is being run at zero forward speed.
Duct integral, I, decreases with increasing sweepback due to inclination of the transverse plane or effective
entry plane which is seen when the duct first becomes fully closed and taken from the rearmost point of
the lip.
As sweepback is decreased, a progressively larger proportion of the pre-entry retardation occurs ahead of
the sweptback surfaces, which is taken into account by decreasing the value of k approx. linearly with
sweptback angle. The value of k shown in the table is compound of the variation of k due to body and
swept surface.
Normally the design point for high speed level flight, determined by the sizing of the entry, lies in the range
of values between 1.4 – 2.0. In this range the theoretical variation complies with experimental results.
Fig 2.4 – theoretical variation loss on leading edge intakes at various angles of sweep.
The above is the results obtained from theoretically calculating the Total pressure loss from equation 2.6. The Matlab
code for calculation is attached in the appendix 2.1
2.4 Pressure recovery characteristics
a. A comparison with measured loss coefficient with that calculated from equation 2.6 shows reveals the
extent of lip separation at low speeds and that of pre-entry separation at high speeds.
b. In terms of practical design, the entry area Ac is an exploitable variable so we cannot use qc for non-
dimensionalyzing the total pressure. It is usual to express the loss in terms of free-stream dynamic head
and make use of recovery factor ƞσi.
2
2
c FdFd Fa Fa
c
P A A ICIC JC JC
q A A
So that
1i
P
q
21 1i
FdFa
P ICJC
q
c. If there were no approach surface, then ƞσi would tend to value 1.0 as is increased to . This will be the
case of a pitot tube registering free stream total pressure.
2.5 Plenum chambers
The analysis of the effect of installing an engine downstream of a sudden enlargement in duct cross-sectional area.
Due to practical space restrictions:
i. It is not possible to diffuse efficiently to the full cross sectional area of plenum chamber so sudden
enlargement is done.
ii. Velocities within the chamber itself are by no means negligible.
plenum Fig 2.5 plenum chamber installation
chamber
Intake duct
c d f
Pressure loss at a sudden enlargement with initially uniform flow - 2 2
1 2 1 1d d d d
d f f f f
P A A A A
q A A A A
Pressure loss in terms of dynamic head at entry – 22
1c d
c d f
P A A
q A A
(2.7)
It can be seen that a modest reduction in effective area Ad can result in large increase in theoretical loss at the
enlargement.
Although plenum chambers intakes are unlikely to come up in future jets but a general knowledge of the effects of
sudden enlargement can be helpful in designing other ducted systems.
2.6 Propeller turbines
Because of their potentially good fuel economy, propeller turbine engines are always in interest of designers.
General intake arrangement is that of an annular intake located directly behind the propeller.
There is a large additional loss as compared to the direct intakes, some 15% of free stream dynamic head at typical
model Reynolds number. Wind tunnel testing shows that the additional loss is attributed to the presence of propeller
blade roots ahead of the intake. The flow over the roots is complicated in character due to large thickness/chord ratio
and the action of centrifugal forces on the boundary layer. An approx. formula for blade root loss:
0.62
bP Nt
q r
where, N – number of blades, t – profile thickness, r – radius of rotation of the section
In annular intake flow when pre-entry flow separation occurs it does so in patches rather than uniformly round the
circumference. So there is a limiting value of above which circumferentially uniform flow cannot exist.
lim 1 1.8A
LR ; where, A – annular entry area, L – length of projecting spinner and the hub, R is its max. radius.
A breakdown of uniformity in the intake has a potentially detrimental effect both on total pressure recovery and also
on engine compatibility.
3. TRANSONIC EFFECTS IN PRE-ENTRY FLOW
3.1 Initial Expectations
M<1 M>1 M<1
a) Subsonic free stream b) supersonic free stream
Fig.3.1 Flow pattern in intake with forward external surface
Total losses in a transonic flow are approach and duct losses as discussed earlier corrected for compressibility effects
in approach loss, together with loss due to shock.
Modified frictional loss equations:
c c
Fd Fa
c
P AIC JC
q A
(3.1)
Or alternatively
2c
Fd Fa
c c
P A AIC JC
q A A
(3.2)
Shock losses:
1
1 ( 1)
22 1
1 2 21 1
1*
1 21 1 1 ( 1)
1 1
P M
PM M
(3.3)
Ps, shock loss is given by the following equation:
2
1
1sP P
P P
(3.4)
From here onwards it is convenient to express pressure loss in terms of total absolute pressure P, so the frictional
loss equation needs to be converted in terms of P by the following relation:
12
2
2 11
2
PM
q M
(3.5)
Shock Front
Total intake loss for this case would be
d a sP P P P (3.6)
3.2 Experiments of Davis et al. (1948)
A comparison between the measured pressure recoveries and those estimated according to equation 3.6 were made.
The very large discrepancy, the measured losses being twice and three times the calculated ones, show that the present
model is incomplete and a major source of total pressure loss exists in addition to those already postulated. Such a
loss can occur only as a result of large scale turbulence associated with flow separation.
3.3 The real nature of pre-entry flow
For subsonic free stream mach number, reducing the flow ratio, Ac/A, imposes a pre-entry pressure gradient, the
boundary layer at entry thickens to a point distorts and then, beyond a certain decrease in flow ratio, boundary layer
separates. Further reduction of flow ratio results in forward movement of separation point in the pre-entry field.
For supersonic free stream mach number a similar characteristics are observed but with two important differences,
a) at flow ratio 1.0 a normal shock wave is sitting at the entry plane, b) the critical flow ratio for boundary layer
separation is now much higher which means that separation will occur much easily.
3.4 Pressure Coefficient at separation
Gadd (1953) obtained an expression for the pressure coefficient at separation based on the assumption that the
separation pressure is that pressure which is just sufficient to bring to rest without friction , the fluid at the knee of the
normal turbulent velocity profile.
Gadd’s formula:
12
22
11
2 2 11
1 .642
pm
MC
MM
(3.7)
(Cpm)incompressible=.36 putting, M=0 and =infinity
This expression may be taken for applying at boundary layer passing through a sudden jump, or infinite pressure
gradient. But in a finite pressure gradient, some energy is entrained into the boundary layer from the main stream in
the distance over which the gradient acts. As the boundary layer distorts, the friction at the surface falls towards zero
and hence an energy surplus is available for increasing the pressure rise.
So,
Cps=Cpm + pressure recovery due to energy fed to BL, Cp
After the derivation we get the following expression for separation pressure coefficient,
( 1)
ps pm
f MC C
J
Where, = gc/ga; J – corrected position ratio as used earlier
4. LIP SEPARATION AND TRANSONIC THROAT FLOW
Pressure loss or recovery characteristics combine the effects of two flow parameters – the mean intake throat
Mach number Mt (which controls compressibility effects) and the position of the dividing (stagnation) streamline at
the cowl lip for a given ratio A∞/Ac (which controls whether or not there is lip separation). The following two graphs
(Fig. 4.1) illustrate how the pressure recovery characteristic is related to Mt and M∞.
4.1 Subsonic diffuser
Aircraft intakes often have ducts with varying shape of cross-section, usually in the shape of an S-bend. Also, in
cases where not much length is available, wall curvature may have to be at a large rate that causes local flow
separation. The resulting additional losses (i.e. in addition to the calculated skin friction losses) can be determined
by fitting the duct with a bellmouth entry and performing a suction test or from computers using Navier-Stokes
equations.
Fig. 4.2 shows the increase in pressure loss with increase of mean throat Mach number, increase of duct length
and changing cross-sectional shape and curving of the duct. When Mt exceeds 0.6 or 0.7, Mach numbers can locally
be transonic at the walls in the region near the throat. In case of curved ducts, there can be some flow separation and
then reattachment. Thus, the loss increases significantly beyond a mean throat Mach number of around 0.6.
In S-bend ducts, the first bend is generally the major source of total pressure loss. Total pressure loss for a given
Mt, length of duct and final bend radius can be considerably reduced by decreasing the amount of turning in the first
bend.
Figure 4.1 Interrelationship between pressure recovery characteristics and Mt and M∞ for axisymmetric pitot intakes having (a) sharp and (b) elliptic lip.1
4.2 Intake plus subsonic diffuser
Instead of performing a bellmouth test, one could obtain a similar result from measurements in an intake plus duct
if A∞/Ac < 1. This condition being satisfied would mean the stagnation streamline being on the lip’s underside,
which would in turn mean the loss of total pressure being only due to skin friction, which for a pitot intake is
typically a function of only Mt for M∞ between 0.6 and 2. Mt can be found from
*
*
t
ct
c t
PA
A PA
AAA
A A
(4.1)
where Pt/P∞ = 1 for subsonic speeds, and it is the ratio of total pressures across a normal shock (NS) for supersonic
speeds. Fig. 4.3 shows a typical plot of total pressure loss, which is defined as
1 at subsonic speeds
and at supersonic speeds
NS
f
f
PP
P P
PP
P P
(4.2)
versus mean throat Mach number. The comparison with a bellmouth test result shows good agreement for values
Connors worked out for axisymmetric intakes, developed the graphical method for optimum shock system. Design
criterion can be decided with graph in Fig 3.16.
Fig 5.15 Shock Pressure Recovery of double wedge intake1
Fig 5.16 Shock Pressure Recovery with varying M∞ and no of shock1
Fig 5.14 Double wedge intake
δ1
δ2 β1 β2
5.4 Isentropic Compression
Let us first derive the Pradtl-Mayer relation for expansion fan and then we can relate it with isentropic compression.
Tangential velocity remain constant, hence
Vt,upstream=Vt,downstream (5.16)
cos( ) ( )cos( )v v dv d (5.17)
sin( )
cos( )
dvd
v
(5.18)
Also, 1
sin( )M
(5.19)
2
1
1
dvd
v M
(5.20)
Now, let us relate v and M using relation v=Ma
dv dM da
v M a (5.21)
Using a RT (5.22)
We get 2
dv dM dT
v M T (5.23)
Since, we are assuming isentropic flow, hence stagnation temperature will not change and relation between T0 and
M is given by
2
0
1(1 )
2T T M
(5.24)
Differentiating the given equation we get
2
0
20
1(1 )
201
(1 )2
d MdT dT
T TT M
(5.25)
Now substituting the value from equation (5.25) into equation (5.23)
2
2
1( )
21
(1 )2
Mdv dM dM
v M MM
(5.26)
Substituting (5.26) into (5.20),
2
2
1
1(1 )
2
M dMd
MM
(5.27)
dM is change in mach number associated with dʋ change in turn angle. δ=ʋ2- ʋ1
which gives
Fig 5.17 Flow over wedge Schematics v
vt
vn vt
vn+dvn
v+dv
dʋ
µ
µ+dʋ
2
1
2
2 12
( 1)
(
1)1
2
M
M
M dM
MM
(5.28)
2
1
2 2
2 1
1 1arctan( ( 1) arctan( 1) |
1 1
M
MM M
(5.29)
Now, if we try to analyze the above derived equation, we would find that the equation can be equally valid for
compression (wedge problem), but instead all expansion would converge to form one oblique shock. This method is
also known as reversed Prandtl Mayer Expansion. With the help of above equation, we can find the maximum
deflection angle by which flow can be turned for maximum pressure recovery, i.e. when M2=1, hence
2 21 1arctan( ( 1) arctan(
1
1 1)M M
(5.30)
5.5 Flow Starting Problem
Let us consider the duct problem attempted in Section 5.1, with difference that duct initially contract and then
increase in area. We have already know with zero exit area, shock wave stand infront of duct and with smaller opening,
shock move towards entry with subsonic flow throughout the duct. There are now two cases, either before shock reach
to the entry, Mt becomes unity or shock reaches entry before shock in the throat.
(a) Shock reaches entry before with Mt<1.This attachment condition implies that A∞/Ac=1. If we further increase
the exit area, shock moves inside duct, but we know that shock can be made only stable inside duct diverging
area (Fig 5.19). An intake in this case is said to started in supersonic sense.
(b) Mt=1 before the shock reaches entry plane (Fig 5.20). If the duct throat area chokes, then further exit area
information cannot cross the throat, hence a second shock (Fig 5.21) will develop in system without having any
change in first shock. Finally, flow will be supersonic from forward detached shock to throat and intake in this
condition is said to be unstarted.
Mt<1
Fig 5.18 Shock reaches entry plane with Mt<1
Mt>1
Fig 5.19 Shock inside duct and flow started
Mt=1
Fig 5.20 Shock forward of entry plane with
Mt=1
Mt=1
Fig 5.21 Forward detached shock and second shock
with flow unstarted
5.6 Limiting Contraction Ratio
If we combine case (a) and case (b) to find limiting value of contraction ratio (At/Ac), then the forward shock would
have to reach entry plane at same time, when flow at throat reaches Mt=1. If ratio is less than limiting value, than case
(b) would be applied and flow would not be started. Mw ∞ being the subsonic mach number after shock.
( 1)/2( 1)
w 2
w
* 1 ( ) M ( )
( 1) M 2wMt
c
AA
AA
(5.31)
Substituting the value of Mw ∞ from normal shock equation (5.3), we get
1/2 1/( 1)
lim 2 2
1 2 2 1( )
1 ( 1) 1 ( 1)
t
c
A
A M M
(5.32)
The graph Fig 5.22 of limiting contraction ratio has been drawn with the help of equation (5.32) and of A*/A from
equation (5.31) with matlab code stated in Appendix 5.5. It seems that χ tends to 0.6 as M∞ tends to ∞. Also the
idealized full internal sonic area relation has been plotted in same graph.
Consider Fig 5.19, consider normal shock has been swallowed up and intake has been started, if we reduce the exit
area, then back pressure will increase causing the shock to move forward toward the throat and hence decreasing its
mach number, hence maximum pressure recovery can be done at throat will stable shock.
APPENDICES
Appendix 2.1 function loss() clear all close all clc u=0:0.01:3; ICFd=[0.14 0.11 0.08 0.05]; JCFa=[0.010 0.021 0.033 0.054]; axes('FontSize', 18) hold all
A*/A
------------- ᵡ
Fig 5.22 Limiting Contraction Ratio and A*/A vs M∞
for i=1:4 ploss(:, i) = ICFd(i) + JCFa(i).*u.^3; plot(u, ploss(:, i), '--') end xlabel('\mu', 'FontSize', 18) ylabel('\DeltaP/qc', 'FontSize', 18) title('Loss vs \mu', 'FontSize', 20) legend('0 degree', '30 degrees', '40 degrees', '52 degrees') end
Appendix 4.1 function lip() clear all close all clc g = 1.4; M1 = 0:0.001:0.7; Minf = [0.115 0.166 0.237 0.330]; t1 = 1+(g-1)*M1.^2/2; tinf = 1+(g-1)*Minf.^2/2; l1 = 1-(t1(91:701)/tinf(1)).^((g+1)/(2*g-2))./((1+g*M1(91:701).^2)
'M_i_n_f_i_n_i_t_y=0.237', 'M_i_n_f_i_n_i_t_y=0.330') end
Appendix 5.1 function M1_net=normalshock(Pr_net) %Pr_net=pressure recovery for which system need to be designed M1=linspace(1,4,100); r=1.4; pr=zeros(1,100);%static pressure ratio pr=p2/p1 Pr=zeros(1,100);%stagnation pressure ratio Pr=P2/P1 M2=zeros(1,100); for i=1:100 pr(i)=1+(2*r/(r+1)*(M1(i)^2-1)); M2(i)=sqrt((1+(r-1)/2*M1(i)^2)/(r*M1(i)^2-(r-1)/2)); Pr(i)=((1+(r-1)/2*M2(i)^2)^(r/(r-1)))*pr(i)/((1+(r-1)/2*M1(i)^2)^(r/(r-
1))); end M1_net=1.2; %initial guess M1_net=fsolve(@(M1)mach1(M1,Pr_net),M1_net); plot([1 4],[Pr_net Pr_net],'--cyan',[M1_net M1_net],[0 1],'--r') hold on plot(M1,M2,M1,Pr) xlabel('M1')
ylabel('M2 & P2/P1') grid end
Appendix 5.2 function fun=mach1(M1,Pr) r=1.4; fun=Pr-((1+(r-1)/2*(sqrt((1+(r-1)/2*M1^2)/(r*M1^2-(r-1)/2)))^2)^(r/(r-
1)))*(1+(2*r/(r+1)*(M1^2-1)))/((1+(r-1)/2*M1^2)^(r/(r-1))); end
Appendix 5.3 function Pr=obliqueshock() n1=100; th1=linspace(.1,60,n1); n2=10; M1=linspace(1.6,3,n2); th2(1:n1)=th1(1:n1)*3.14/180; r=1.4; b1=zeros(n2,n1); Pr=zeros(n2,n1); for i=1:n2 m1=M1(i); for j=1:n1 th=th2(j); b1(i,j)=fsolve(@(b)betashock(b,m1,th),th); b=b1(i,j); m2=sqrt((1+(m1^2*(sin(b)^2)*(r-1)/2))/(sin(b-
%P3/P2 Pr(i,j)=Pr1*Pr2; %P3/P1 end for i=1:n2 plot(th1(1:1.6*M1(i)*10),Pr(i,1:1.6*M1(i)*10)) hold on end plot([0 30],[.96 .96],'--r') xlabel('Wedge Angle') ylabel('Ps/P1') grid
Appendix 5.4 function fun=betashock(b,m1,th) r=1.4; fun=tan(th)-2*cot(b)*(m1^2*(sin(b))^2-1)/(m1^2*(r+cos(2*b))+2); end
Appendix 5.5 function isentropic() n2=100;
M1=linspace(1,5,n2); M11=linspace(1.5,5,n2); r=1.4; v1=zeros(1,n2); v2=zeros(1,n2); for i=1:n2 m1=M1(i); v1(i)=sqrt((r+1)/(r-1))*atan(sqrt((r+1)/(r-1)*(m1^2-1)))-atan(sqrt(m1^2-