The Society shall not be responsible for state-

ments or opinions advanced in papers or in dis-

cussion at meetings of the Society or of its
Divisions or Sections, or printed in its publications.

Discussion is printed only if the paper is published

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

in an ASME journal or Proceedingi.
$1.50 PER COPY
Released for general publication upon presentation
75C TO ASME MEMBERS
Copyright © 1967 by ASME

Jet Engine Starters, Cartridge-Pneumatic

J. A. ANDERSON
Group Engineer,
Cartridge-Pneumatic Starters,
Sundstrand Aviation, A Division
of Sundstrand Corporation,
Rockford, Ill.

C. R. GALASINSKI
Project Engineer,
Cartridge-Pneumatic Starters,
Sundstrand Aviation, A Division
of Sundstrand Corporation,
Rockford, Ill.

A typical dual-mode (cartridge-pneumatic) jet engine starter is discussed with respect

to its basic design and functional purpose. The establishment of starter output require-
ments for satisfactory engine starting is presented as an aid to understanding the de-
sign problem encountered for a jet engine starter. Discussion of energy conversion and
particular design parameters concentrates on the cartridge mode of operation and
the handling of the high-temperature, high-pressure gas produced by the starter cart-
ridge.

Contributed by the Gas Turbine Division for presentation at the Gas Turbine Conference
and Products Show, Houston, Tex., March 5-9, 1967, of The American Society of Mechanical
Engineers. Manuscript received at ASME Headquarters, January 27, 1967.
Copies will be available until January 1, 1968.

THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS, UNITED ENGINEERING CENTER, 345 EAST 47th STREET, NEW YORK, N.Y. 10017
Jet Engine Starters, Cartridge-Pneumatic
J. A. ANDERSON C. R. GALASINSKI

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

1 INTRODUCTION There are two particular utilization advan-
tages to the approach of the cartridge-pneumatic
It is the purpose of this paper to present starters to jet engine starting. The first utili-
a discussion of the significant parameters which zation advantage is the self-sufficient starting
must be considered when sizing or designing a car- capability. This means that when operating in
tridge-pneumatic type of starter. Special atten- the cartridge mode, it is possible for an aircraft
tion is given to the design areas which are unique equipped with cartridge-pneumatic starters to com-
to this particular type of device. plete its start cycle without additional ground
To assist and orient the reader who may not support equipment.
be familiar with jet engine starters, an initial The advantages of this are obvious, in that
section, 2, which describes a typical cartridge- aircraft can be dispersed or operated out of re-
pneumatic starter and its functional purpose, is mote or inconvenient areas where ground support
included. equipment either cannot be made available or, at
any rate, does not have to be made available. It
2 FUNCTIONAL PURPOSE AND DESCRIPTION should be pointed out here that the combination
OF A TYPICAL UNIT starter device does not have to be operated in the
cartridge mode for all starts. This means that
The primary purpose of all jet engine when ground support equipment is available, par-
starters is to provide a torque level sufficient ticularly for maintenance starting, the pneumatic
to accelerate the main engine rotor from zero rpm mode can be utilized. This extends the service
to engine light-off, and to assist the engine to a life of the starter and reduces the actual start-
speed at which it alone can sustain the accelera- ing costs. The second utilization advantage of
tion to engine idle without excessive internal cartridge pneumatic starters is the alert or
temperatures. The unit is normally mounted di- quick-start capability that can be provided.
rectly on the engine accessory gearbox, which is Since the aircraft can be started without any at-
mechanically connected to the engine rotor. Air- tached ground support equipment, the aircraft can
frame facilities are provided for pneumatic inlet be available for flight just as soon as the en-
ducting, electrical power for cartridge ignition, gines are brought up to speed. There is no time
accessory exhaust ducting, and access for car- lost in the disconnecting of ground support equip-
tridge loading. ment and coordination of this starting operation
with a ground crew.
Fig.l shows a simple schematic of a combina-
tion cartridge-pneumatic starter. It can be seen
that this starter arrangement is made up of a
cartridge gas generating system, a turbine or
CARTRIDGE power-conversion arrangement, and a reduction gear
box and disconnector arrangement. The power from
GROUND CART
either the cartridge gas generator during a car-
tridge-mode start or from a ground cart or air-
ENGINE INTERBLEED I
craft interbleed system is transferred to appro-
4
11111111P I priately designed nozzles, so that the same tur-
rPOWER POWER REDUCTION GEAR - 1- bine is used for power conversion in either mode.
GENERATION CONVERSION BOX AND DISCONNECT A description of a typical cartridge-pneu-
Fig. 1 Cartridge-pneumatic starter schematic matic starter is shown in Figs.2 and 3 by a sche-
matic of the functional parts and a cutaway view.
Fig.4 shows a list of the leading particulars of
this one starter example.

1
 Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022
13

Fig.2 Schematic of functional starter components

1 Cartridge 10 Overboard exhaust connector 19 Flyweight

2 Breech cap 11 Exhaust from turbine and fan 20 Switch actuating rod
3 Breech handle 12 Relief valve 21 Switch
4 Connector 13 Compressed-air inlet 22 Adjusting screw
5 Ignition contact 14 Aerodynamic-braking fan 23 Gearbox vent
6 Ground clip 15 Air inlet for braking fan 24 Spline shaft
7 Hot-gas nozzles 16 Fan exhaust ring 25 Oil slinger
8 Turbine rotor 17 Gearshaft 26 Oil sump
9 Turbine exhaust ring 18 Overrunning sprag clutch 27 Magnetic plug

3 ESTABLISHMENT AND DISCUSSION OF SIGNIFICANT drag-torque loading, and the desired aircraft per-
DESIGN PARAMETERS formance. The starter output requirements may be
presented in specific terms, or in terms of en-
The sequence of this section presents first gine or aircraft operational requirements.
a discussion of the parameters which determine In specific terms, the requirements are
the starter output requirements; then in the sub- generally presented as an actual "starter torque
sequent sections the energy conversion is followed versus speed envelope," as shown in Fig.5. To
through the starter from the cartridge to the noz- present the requirements in this form, the air-
zle, to the turbine rotor, and to the output frame manufacturer must perform the necessary cal-
shaft. In the last section, special items are culations to assure that the minimum torque curve
covered which are unique to cartridge starters, will result in satisfactory engine starts.
such as hot-gas handling, speed control, and In many cases, however, the starter output
safety devices. is presented in terms of the engine starting re-
quirements and aircraft operational requirements.
A Establishment of Starter Output Requirements In this form the airframe manufacturer takes the
The starter output requirements are normally engine drag-torque curve from the engine manufac-
established by the airframe manufacturer and are turer and adds the drag-torque loading for the
based upon the engine requirements, the accessory specific hydraulic and electrical accessories.

2
 Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022
1. Electrical Connector 10. Containment Clamp
2. Switch 11. Turbine Rotor
3. Switch Actuating Rod 12. Aerodynamic Braking Fan
4. Lubrication Tube 13. Breech Cap Assembly
5. Output Clutch Race 14, Breech Cap Chamber
6. Output Spline Shaft Assembly 15. Pneumatic Inlet
7. Adapter (Installation Part) 16. Ignition Connector
8. "V" Band Coupling (Installation Part) 17. Transfer Tube
9. Gearshaft
Fig. 3 Cutaway view of cartridge-pneumatic starter

The resultant curve may look like Fig.6. The air- tion is given to the rate of acceleration to
craft operational requirements may also dictate a light-off because of the viscous drag loading of
maximum time-to-idle curve such as Fig.7. the engine and accessories. The minimum assist or
The significance of the curves and limits of cutout speed for the starter is determined by the
Fig.6 is as follows. The maximum torque limit is minimum positive torque above which the engine can
determined by the load-carrying strength of the accelerate itself uniformly to idle.
accessory gearbox and engine accessory power Given the information in Figs.6 and 7, the
train. The -65, +59, and +130 F drag curves are starter designer can begin to approximate the
representative of the typical engine plus acces- shape of the starter output curve which will best
sory drag torque at that specific ambient tempera- satisfy all the requirements of minimum engine
ture. At the -65 F condition, special considera- assist and time to idle. The resultant time to

3
 Weight ............................................................................ 61 lbs. 6 oz.
Operational Temperature Range ........................ -65 to +160 ° F.
Maximum Operational Altitude ........................ 6000 feet
Operational Attitude ................................................ Horizontal +10 degrees
Output Rotation
(when facing output spline) ................................ Counterclockwise
Maximum Output Shaft Torque ............................ 680 pound feet
Output Spline Shear Section ................................ 800 to 900 pound feet
Pneumatic Shut-Off Actuation

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

(output rpm) ............................................................ 2970 + 100 rpm
External Disengagement at Output Spline
Shaft (output rpmp) ............................................ 3000 rpm
Maximum Turbine Operational Speed . . 67, 500 rpm
Cartridge Required .................................................... Air Force Type MXU-4/A
or MXU-4 A
Electrical Requirement for Ignition ................ 18 to 30 volts dc at 1 ampere
Relief Valve Actuation Point ................................ 700 psi
Gear Ratio 14 88 to 1
Pneumatic Requirement
Source ........................................................................ Compressed air from MA-1A
pneumatic ground cart or
equivalent
Maximum Pressure ............................................ 60 psia
Permissible Frequency of Use
Cartridge Starts per Hour ................................ 2
Minimum Time Between Starts on
Aircraft (minutes) .................................... 5
Minimum Time Between Starts
During Testing (minutes) .................... 30
Maximum Continuous Motoring ............................ 10 minutes
Minimum Waiting Time before Removal
of Spent Cartridge* ............................................ 1 minute
Minimum Waiting Time before Removal
of Hang Fire Cartridge* .................................... 15 minutes
Oil Requirements
Specification ............................................................ MIL- L-7808
Maximum Oil Capacity .................................... 275 cc(9. 3 ounces)
*Assumes use of asbestos gloves
Fig. 4 Table of leading starter particulars

reach engine idle (t) is derived from the follow- where

ing relationship. T = net accelerating torque reflected to
n
the starter mounting pad
I = total inertia reflected to the starter
T = I a (lb/ft) (1) T
n T output shaft: l e = engine, I A = acces-
sory, I s = starter
I = I + I + I (slug-ft2) (2)
T e A s T = steady-state starter torque at the
s
1- starter mounting pad: Te = engine
T = Ts - (T e + T Acc ) (1b/ft) (3)
n torque, T Acc = accessory torque

4
 LL +130

I +59
LI)

z
ka.i -65
6-5

30 40
STARTER OUTPUT SPEED IN) TIME TO IDLE (SEC .1

Fig. 5 Starter torque versus speed envelope Fig. 7 Engine starting requirements (time to idle)

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

-65°F
4000
+60°F
MAX ALLOWABLE STARTER TORQUE +160°F 3000

2000
-65°F
+59°F
0.15 1500
+130°F

010 1000
009
STARTER SPEED
§ 0.08
2 0.07
MIN ASSIST 0 06
100 LB.-FT 0 05

004

0.03
ENGINE LITE- OFF = 200 RPM + 8 TO 10 SEC 200 300 500 1000 2000 3000 500

CHAMBER PRESSURE, PSI

INERTIA ID OF ENGINE
PLUS ACCESSORIES AT
2 EFFECTIVE GRAIN BURNING THICKNESS = 119 IN. 2
STARTER PAD =IXISLUG-FT.
EFFECTIVE GRAIN BURNING SURFACE IA, I= 129 8 IN.

Fig. 6 Engine starting requirements (torque versus speed) A

t
= EFFECTIVE NOZZLE AREA

Fig. 8 MXU-4/A Cartridge burn rate and K n versus chamber

pressure and grain temperature (fixed nozzle system) (Courtesy
of Olin. Mathieson Chemical Corporation)

a = acceleration reflected to the starter B Starter Cartridge Input Energy Available

mounting pad The energy source for self-sufficient start-
N = starter output speed ing of an aircraft utilizing a cartridge-pneumatic
t = time to idle starter is a starter cartridge such as the MXU-4/A
or MXU-4A/A. The primary constituent of these
Then, from equation (1):
cartridges is an 8-lb grain composed of pellets of
an ammonium nitrate base propellant homogenously
dn
T n = I T &t (lb/ft) ( 4) suspended in an appropriate combustible binder.
The design of the grain is such that the burn
rate, or rate of consumption of the grain, pro-
t - IT ,r
idle
0
(sec) (5) duces a relatively constant mass rate of gas flow
having a characteristic temperature of approxi-
The actual calculation of time to idle using mately 2400 R, and composed of CH4, CO 2 , H 2 , H2 0,
equation (5) is performed by taking AN increments, N 2, and Na CO
2 3'
using the average torque for each increment, and It is also characteristic of ammonium ni-
determining the At for each increment; i.e.: trate base propellants to exhibit a burn rate
which is directly proportional to a function of
I x 2/t/60x (4h.RPM) (sec) (6) the conditioned temperature of the grain and the
t
Tnet operating chamber pressure. Graphical representa-
time to idle (t) =1 0 t (sec) (7) tions of cartridge chamber pressure versus time,
burn rate and K versus grain temperature and
n
The same method of calculation is used to deter- chamber pressure for the typical MXU-4/A cartridge
mine the engine speed at cartridge burnout. are shown in Figs.8 and 9. The term K is the
n

5
 1 ,1
1500

Lailemarmos 1500

II 111 \
1000

500

0
I t
a_
1500

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

V)
Er)
100 0
k eililiMMEMIN Fig. 10 Cartridge burn characteristics (fixed-nozzle system)

rlillii iirillirili -
cc
a_
re
-N
50 tures of -65 and +160 F while using a valve to
produce a variable orifice. Fig.11 also shows

IW= -- 1ENE1MIN \\ that, with the variable orifice design, the re-
duced cartridge pressure results in a slower burn
rate and correspondingly longer cartridge burn

ril —65°F ■•■MI■LUIll

500
time for the hot-day conditions. This resultant
1 characteristic is beneficial in starting engines
I6 2 28
which are slow to light-off.
0 8 I2
THE,SECONDS
The available energy, or available adiabatic
head (H ad ), produced by the cartridge may be cal-
Fig. 9 MXU-4A Cartridge chamber pressure versus time culated from published values of the cartridge
(fixed-nozzle system) (Courtesy of Olin Mathieson Chem- specific heat ratio (k), specific gas constant
ical Corporation) (R), the flame temperature (T p ), and the appropri-
ate design operating pressure (Po) and exhaust
ratio of nozzle-throat area to the burning surface exit pressure (Pe). Thus
of the cartridge grain, and therefore is use-
ful in predicting chamber pressures for a particu- H
ad = KR T [ 1-(pe ) (ft) (8)
lar nozzle system. K-1 Po
It is evident from Figs.8 and 9 that, for a
fixed-nozzle size, the increased burn rate of a
+160 F cartridge will result in an increased For the typical MXU-4A and MXU-4A/A cartridge gas:
chamber pressure, which in turn results in an ad-
ditional increase in the burn rate of the grain R = 81
until stabilization is reached. Typical chamber K = 1.27
pressure versus time curves for cartridge tempera-
tures of -65 and +160 F, with a fixed-nozzle size, T =2400 ° R
are shown in Fig.10. However, since the output
torque of the starter is a function of the pres- To simplify equation (8), let
sure ratio across the nozzle, the magnitude of
this chamber pressure variation with temperature K-1
produces excessive torque output at the high ambi- Y = 1-( Pe) K (9)
Po
ent temperatures required.
The starter described in Section 2 therefore
utilizes a highly pressure-sensitive valve, which Then, from Fig.12, the value of Y can be deter-
produces a variable-nozzle bypass orifice. The mined for any Pe/Po with K = 1.2 to 1.4. There-
resultant increase in effective nozzle area re- fore, equation (8) may be expressed as
duces the chamber pressure below the fixed-nozzle
condition, such that the pressure difference be- H
ad = K RT,Y (ft) (10)
tween the hot and cold cartridge is minimized. K-1
The maximum starter output torque and, subse-
quently, the impact torques sustained by the en- The spouting velocity (Co), or maximum
gine gearbox can then be controlled within a more theoretical gas velocity which can be obtained at
desirable range. Fig.11 shows the typical chamber a nozzle exit from the available energy, can then
pressure versus time curves for cartridge tempera- be determined from

6
 825
750

16 21
TIME - SECONDS

Fig. 11 Cartridge burn characteristics (variable-nozzle system) 05

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

P
E

0
1/2 1/2
C o = ( 2g H ad ) or (2g kRT p Y)
k-1 01

(ft/sec)(11)
005

The energy available from the cartridge may also

be calculated in terms of gas horsepower (GHP)
from the mass rate of cartridge gas flow (W f ) and
the adiabatic head (H ad ), where W f is a function
001
of the grain burn rate (r b ) at the desired operat-
ing pressure and temperature, the grain burning
area (A s ), and the grain density (p ). Thus 0005

r, from Fig.9
W = rb As (12)
f
A s = 129.8 in 2 (typical)
jOl p = .053 #/in 3
0001
4
0
w H
GHP - 550
f ad (HP) (13)
Fig. 12 Pressure ratio versus Y
Approximate values of the above properties for a
typical MXU-4A cartridge at nominal operating con- diverging nozzles requires that the Mach number
ditions of +60 F, 865 psia, and exhaust pressure (M) at the nozzle throat must be unity, and that
of 18 psia are the pressure ratios P e /P o must be less than the
H
critical pressure ratio P*/P o required for M = 1
ad :Z.- 506 x 10 3 FT at the throat, where P* is the pressure at the
C 3
o C:= 5.7 x 10 FT/sec throat, and the critical pressure ratio P*/P o is
defined by
f .52 1,13 m /sec

481 HP k
GHP *
P ( 2 ) K -1 (14)
Po k+1

It should also be noted that the actual de-

C Starter Nozzles and Nozzle Energy Conversion sign value of P o must not only be consistent with
The first stage in the transformation of the the requirements for supersonic isentropic flow,
heat energy produced by the cartridge to usable but must also be of a sufficient magnitude to pro-
work is the conversion to kinetic energy by a noz- duce the required starter performance with the
zle system. The starter accomplishes this energy final nozzle and turbine design incorporated in
conversion by expansion of the gas through super- the starter. Experience has shown that values of
sonic converging-diverging nozzles. This type of P o in the range of 800 to 1000 psig are generally
nozzle is used because of its characteristically sufficient.
high efficiency for high pressure ratios P o /P e . The conditions of the gas at the nozzle exit
Before defining the energy conversion by the may be determined from the following pressure,
nozzles, it should be recalled that supersonic temperature, and density relationships for isen-
isentropic flow of a gas through converging- tropic flow:

7
 12
le max = (Kg) /
k -1 K+1 (21)
Po k = To (FTC/sec)
or (Po ) Tr— K-+1N 2(K-1)
15; -e (15) k 2 I--

For the typical cartridge gas

7 = .Po (16)
e 'doe K 1.27

Max. 3.78
where p is the gas density.
As noted in section B, the maximum theoreti-
Therefore
cal exit velocity of the gas at the nozzle exit

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

may be determined from
W f = A* P o max(lbm/sec) (22)
(RT 0 )
12 (ft/sec) (17)
Co = ( 2°48- 6 ) /

The actual nozzle exit velocity, however, is

D Starter Turbine Rotor and Turbine Energy
dependent on the nozzle velocity coefficient (k v ),
Conversion
which is the ratio of the actual exit velocity
The second stage in the transformation of
(V ) to the theoretical exit velocity with isen-
1 the cartridge gas energy is the conversion of the
tropic flow and the same exit pressure P e . The
kinetic energy of the gas leaving the nozzle sys-
value of k v for supersonic nozzles having straight
tem to usable shaft power by a means of a single-
axes has been found from experience to generally
stage impulse turbine.
equal 0.96.
The absolute blade speed used for the design
The actual nozzle exit velocity is, there- of a starter turbine is not necessarily the speed
fore: at which maximum efficiency is obtained, because
the starter must be designed to yield a maximum
V1 = .96 C o(Ft/Sec) ( 1 8) output torque at the most critical starter output
speed. The design speed used for the blading de-
To summarize the conversion of the heat en- sign, and therefore the entire starter system de-
ergy of the gas to kinetic energy, it can be sign, is therefore the one corresponding to the
stated that the nozzles transform the low- starter output speed at which the maximum engine
velocity, high-temperature, high-pressure gas pro- drag torque must be exceeded. The actual turbine
duced by the cartridge to a lower static tempera- blading utilized by a cartridge-pneumatic starter
ture, lower static pressure gas having a super- must also be designed on the basis of a compromise
sonic velocity and the same mass rate of flow. between the most practical turbine efficiency for
The nozzle exit W can be determined from both the cartridge and pneumatic-energy sources.
f
the relationships for isentropic flow at the The reason for this is that the mass rate of gas
critical nozzle area A*. flow from the pneumatic-energy source is generally
t'
greater than that produced by the cartridge, and
W
f
= Jo* C* A* ..e e
Fi. 7,
P the size and weight requirements of the starter
assembly necessitate the use of a cascade of con-
C* (KgRT*) 1/2 • (KgR ) 1/2 T o 1/2 verging nozzles for most efficient energy conver-
(T-71) (19)
sion of the pneumatic gases. Therefore, the noz-
zle exit conditions and velocities are signifi-
A* = Kd A t * K d = Nozzle discharge coeff = .96 cantly different for the two energy sources.
Let us,begin the analysis of energy conver-
Therefore: sion by an impulse turbine by defining an impulse
force as one which results from a decrease in
A* P o (Kg) 1/2 magnitude or change in direction of the tangential
Wf = velocity of a fluid.
(RT 0 )1/2 )K+1 (lb m/sec) (20)
"' T7=1) For the ideal impulse blade shown in Fig.13,
if the absolute tangential velocity of the blade
For pressure ratio (Po/Pe) greater than the criti- is U and the absolute velocity of the fluid leav-
cal ratio, the foregoing Wf equation can be sim- ing the nozzles and entering the blade is V 1 , then
plified as follows by using the maximum compressi- the relative velocity of the fluid entering the
bility factor ( y max), where blade (W 1 ) is, vectorially:

8
 VI

W =v - U
I 1
U

VZ

W 2 =-W I =U - V 1

Fig. 13 Schematic of ideal impulse blade

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

Fig. 14 Velocity diagram for simple impulse turbine
W V1
= --.- U (23)
1

Using similar notation, and assuming fric-

tionless contact between the fluid and the blade,
/7/ = r ( dw ( V 1U - V2U
it can be seen that the relative velocity of the dt /
fluid leaving the blade (W 2 ) must be equal in (lb/ft) (28)
magnitude but opposite in direction to W. There-
T = (V1U - v2u)
fore

W2 =--. --a
"' W
1 The analysis of a real impulse turbine is
(24)
U --0.• V1 facilitated by a velocity diagram such as the one
shown in Fig.14, where, in addition to the nomen-
Then, since W2 is opposite in direction to Wi and clature previously defined:
V 1 , it can be seen that the absolute velocity of a = angle of absolute fluid velocity V 1
the fluid leaving the blade (V2) must be leaving nozzle
0 = angle of relative fluid velocity W1
V2 = W 2U ===-1.W1.4-11. U entering blades
(25)
= 2U V1 y = angle of relative fluid velocity W 2
leaving blades
This brief analysis of the ideal impulse 8 = angle of absolute fluid velocity V2
blade clearly shows that the effective tangential leaving blades
velocity of the fluid entering the blade (V 1U ) is V ia and V 2a = resultant axial components of abso-
decreased, resulting in an impulse force. The lute fluid velocities V 1 and V 2 .
force produced by the gas is derived from Newton's Using equation (28) and the relationships
laws of motion, which state that momentum is the
product of the mass of a body and its velocity and V 1U = V1 Cos oCand V2u = V 2 CoSS
that the rate of change of momentum is equal to
the sum of the external forces acting on the body. derived from the velocity diagram of Fig.14, the
Therefore value of T for a real impulse turbine may be bx-
pressed as
F = d (M) = d ( E— (1bf) (26)
dt dt g
= W f r (V 1 Cos0C- V2 Cos ) (lb/ft) (29)
g
The impulse force produced by the tangential ve-
locities of a gas entering and leaving a turbine Since the relationships derived from the analysis
blade passage may therefore be defined as of the ideal impulse blade assume steady flow and
no blade passage losses, it is only necessary to
apply suitable design coefficients for utilization
F
i = )( V
1U V2U) (lb f ) (27)
dt W of those relationships for practical turbine
analysis and design.
Now by noting that the mass rate of fluid flow These design coefficients, which account for
(W f ) is equal to dW/dt, and that torque about a the difference between the relative fluid entrance
fixed axis is equal to the product of the force velocity W i and the relative fluid exit velocity
and the distance from the axis to the point at W cannot be calculated because of the numerous
2'
which the fluid enters the blade (r), the turbine indeterminable factors which effect the blade
torque produced by the fluid may be defined as passage losses. However, from the interpretation

9
 torque, which occurs at zero rotor rpm, may then
be derived from equation (32) by equating k p k i =
kb and setting U = 0. Therefore

Ts
= w f rv, (Cos + k b k d Sin Cos r
2 4 6 8 L0 Sin/
U/v i
(lb/ft) (33)
Fig. 15 Typical torque coefficient versus 11/V 1
(cartridge mode of operation) For practical analysis of starter turbine
systems, it is a common practice to account for

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

all the losses which occur in the nozzle-turbine
and evaluation of extensive testing, a satisfac-
system and the fixed geometric parameters by means
tory correlation between predicted and actual re-
of a stall torque coefficient, designated as
sults has been obtained by assuming that the sig-
tau (
nificant losses occurring in single-stage impulse
From equation (33) let
turbines are represented by a profile-loss coeffi-
cient (k p ), and an incidence-loss coefficient
(ki). The coefficient k p accounts for the losses kb k d Sin of Cosr (34)
T max = Cos a
'

Sinif
due to fluid turbulence, friction, deflection of
the fluid within the passage, curvature of the
The equation for turbine stall torque may then be
blade profile, and the deviation of the actual
expressed as
fluid exit angle y from the geometric blade exit
angle y'. The coefficient k i accounts for the
losses due to turbulence caused by the difference Ts = '
rmax W
fV1 r (lb/ft) (35)
between the entrance angle 13 of the relative fluid
velocity W i and the geometric blade entrance angle
Also, since the turbine torque will be equal
to zero when the tangential blade velocity is
Another coefficient, which is required for
equal to tangential velocity of the gas entering
partial-admission systems only, is the partial-
the blade, V 1 Cos a, the value of T at the corre-
admission factor kd. The value of this coeffi-
sponding rotor velocity must equal zero. However,
cient is a function of the blade profile width
in the actual turbine system, the maximum tangen-
(c), the geometric blade exit angle (y 1 ), the
tial velocity of the gas will never equal V 1 Cosa,
thickness of the blade trailing edge (t e ), and the
but will incur some losses due to the deviation of
arc length of admission (e), and may be calculated
the angle of the gas from the geometric angle of
from
the nozzle.
Therefore
kd = 1 -
E yo.

2 e
-0-t e(30)
U = KV Cos c<.. or ( U = K Cosc.C.
max 1
1 ) max (36)
Therefore, recalling that the effect of the blade
losses is to decrease the magnitude of W2, the
Empirically it has been found that for the starter
actual value of W 2 is defined by
turbine systems, which have a nozzle angle (a) in
the range of 17 to 20 deg, the maximum value of
W2 = k p k i kd (ft/sec) (31)
(U/V 1 ) max is approximately equal to 0.90, which
is also the speed ratio at which 'C equals zero.
Using the relationship of equation (31) and the
Therefore, since turbine torque versus speed
identities
is very nearly a linear function, a curve of T
V 2 Cos S = U - W2 Cos r and W 1 = V1 Sin- versus U/V for a typical starter turbine system
1
Sin/g may be derived as shown in Fig.15.
from Fig.15, equation (29) may be expressed as The simple expression of torque in equation
(35) is very useful in the evaluation of modifica-
T. = WfrV ( Cos kp k i k d Sin 64. Cos - u tions to an existing unit configuration and for
Sin /9 predicting unit performance with intermediate or
new cartridge pressure levels. With a known
(lb/ft) (32)
fixed-nozzle diameter, the value of W f and V 1 can
A simplified equation for the turbine stall be determined from the measured breech pressure.
torque T s , or the maximum obtainable turbine The 'C factor can then be determined from the

10
measured starter output torque, or the torque can
be predicted by a previously determined "t" versus
U/V relationship.
1 600 120
It is apparent that small-diameter impulse
CO HP VS SPEED
turbines must operate at extremely high rotative CO

%EMMEN=
500 100

speeds for most efficient energy conversion of the

w 400 80
high-energy gas produced by a cartridge. There-
fore, since the torque-speed characteristics of
ce
• ■■11E 60
300
VS. SPEED
the turbine will almost never match the torque-
200 • 1111■
3RQUE 40
speed characteristics required for starter output,
a gear ratio must be employed between the turbine 100 F■IIIIM11■■ 20

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

and the starter output shaft. 0
The value of this gear ratio must be a com- 1000 2000 3000 4000
STARTER OUTPUT SPEED — RPM
promise between the gear ratios which will produce
the most desirable starter torque-speed curves for Fig. 16 Typical starter torque and starter horsepower
both the cartridge and pneumatic modes of start- versus output speed (cartridge mode of operation)
ing. The reason for this compromise is that, be-
cause the stall torques resulting from a typical The first problem encountered was that of
Air Force MA-1A type ground cart pneumatic-energy material cracking due to thermal cycling stresses.
source are normally lower than those resulting The selection of materials used to confront this
from a cartridge energy source, the required slope problem was accomplished mostly by experimental
of the torque-speed curve for the pneumatic mode testing with the high-temperature nickel and co-
must be much less than that of the cartridge mode balt-base alloys. Test components were made from
in order to exceed the engine drag torque at all both investment cast and wrought materials. In
output speeds below the required minimum starter most of the applications it was found that the
assist speed. The gear ratios actually used for cobalt-base alloys exhibited a slightly higher re-
starter systems range from 13:1 to 22:1. sistance to thermal cracking, and that the wrought
Utilizing the required gear ratio (G) and material provided a longer endurance life. It was
the corresponding efficiency (% ) of the gearbox also found that the resistance to thermal cracking
G
required, the actual starter output torque may be was greatly increased by the incorporation of the
determined from most practical gradual transition between thin and
thick cross sections, and by utilization of the
most generous radii practical, at all internal
T starter
= TWf r via
g
(lb/ft) (37)
corners. Weld joints were found to be particu-
larly prone to cracking in the heat-affected area
Typical torque versus speed and horsepower versus of the parent material, which was minimized by
speed curves for a starter operating in the car- using equal cross sections for the components at
tridge mode at ambient temperature are shown in the weld joint, whenever possible.
Fig.16. High-temperature corrosion and/or erosion
potential exists in all areas exposed to the flow
E Hot-Gas Handling and Transfer System of the gases. Here again, the wrought cobalt-base
Special attention must be given to the se- materials exhibited the better resistance. It has
lection of the materials used for the starter been found that the cartridge ignition gases con-
components carrying the cartridge gases, in order tain sodium and potassium salts along with various
to provide: degrees of sulfur compounds, depending upon the
1 Sufficient pressure vessel strength at particular ignition design. These elements, in
maximum temperature. combination with a reducing atmosphere, react with
2 Minimum surface temperatures. the metals at high temperature to form nickel and
3 Minimum localized fatigue stresses due chromium sulfides. Also, the chromium reacts with
to high-temperature cycling. the carbon in the reducing atmosphere to form
4 Minimum erosion by the cartridge gas. chromium carbides. Both of these reactions cause
5 Minimum high-temperature corrosion by the the chromium to be depleted as a corrosion-resist-
cartridge gas residues. ant element and allow the nickel or cobalt to
The problems encountered by the designer and later oxidize more rapidly. It is generally found
metallurgist in achieving the desired endurance of that the nickel alloys are impervious to this
the components subjected to contact with the high- corrosion below 1200 F and the cobalt alloys are
temperature cartridge gas (1900+ F) have been good for approximately 200 deg F higher. Diffu-
formidable. sion coatings are helpful in resisting this corro-

11
 I BREECH CHAMBER ASSEMBLY

SWITCH

BREECH CHAMBER PRESSURE VESSEL

INTERNAL GAS SHIELD

INTERMEDIATE RADIATION SHIELD

GAS FLOW LINER TUBE

INSULATION
INSULATION 8 RADIATION Fig. 18 Overspeed protection which diverts gas before nozzle
GAS FLOW

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

INTERNAL (FREE FLOATING)
GAS TUBE
PRESSURE VESSEL TUBE
OUTER SKIN

III NOZZLE BLOCK ASSEMBLY

OUTER SKIN

w‘s mmvi s‘‘‘m s‘

. IN S U L A T I 0 N
ex
r x x
PRESSURE VESSEL
GAS FLOW ► IN RADIATION SHIELD
AND INSULATION
FLOW LINER

SENSING ARRANGEMENT
AND GEARBOX

APPROACH 2

Fig. 17 Schematics of hot-gas transfer components Fig. 19 Overspeed protection which diverts gas after nozzle

sion, and in some areas of the starter where cor- diameter required for sufficient output torque, it
rosion was particularly bad, a chromized diffusion is not practical to design a turbine rotor which
coating has been used with success. can structurally withstand a free-running condi-
The basic design approach used for the tion at tip speeds corresponding to rotor speeds
starter described in section 2 was to use a mate- near 160,000 rpm. It is therefore necessary to
rial which exhibited high corrosion resistance and provide a reliable means of limiting the maximum
good high-temperature strength, as an internal gas rotor speed.
shield and/or gas-transfer member, wherever possi- There are several methods of providing tur-
ble. These internal shield and/or transfer mem- bine overspeed protection in a cartridge starter.
bers are only required to direct the main flow of A few of the devices which have been utilized are:
cartridge gases, and are not required to withstand 1 Diversion of the hot-gas flow so that it
the stresses produced by the gas pressure. There- does not pass through the nozzle when cutout speed
fore, these members are designed to be free-float- is attained.
ing wherever possible, thus minimizing the 2 Diversion of the hot gas between the noz-
stresses due to thermal expansion and internal zle exit and the turbine inlet, either partially
pressure. or completely, when cutout speed is attained.
The structural members, which must actually 3 Modification of the single-stage turbine
withstand the gas pressures, are protected from output torque so that the turbine is aerodynami-
the high-temperature cycling by the use of insula- cally or inherently speed limited within a safe
tion materials, air gaps, and/or shielding between design speed.
it and the gas-carrying member. From Fig.17 it Each of these methods of speed limiting has
can be seen how this design approach is applied to been used with varying degrees of success, and
each area of the starter which transfers the hot there are advantages and disadvantages to each.
gases. Method 1. Fig.18 shows an arrangement
whereby the gas is diverted before reaching the
F Speed Control and Safety Devices nozzle. In this case, it is necessary to have
Because of the high spouting velocity (C o ) some type of speed or acceleration sensing device
of the cartridge gas (5700 fps) and the rotor with a cutout or bypass arrangement in the cir-

12
 FAN

0
0
CARTRIDGE RUBBER SEAL
REQUIRED LEAKAGE GAP

ELECTRIC CONNECTOR AND BREECH LOCK

Fig. 21 Breech locking design

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

REENTRY DUCT

G Breech Electrical Interlock

Fig.21 shows a typical starter cartridge
breech assembly. It includes the basic pressure
vessel in which the cartridge is burned as well
as a joint at which the pressure vessel separates
for the purpose of installing the cartridge.
Fig. 20 Aerodynamic speed control The breech must also include a locking de-
vice for indexing and locking the two halves of
the breech after the cartridge has been installed.
This locking device includes the contact for the
cuitry. The sensing device may include a flyball electrical circuit which is used to ignite the
governor system in conjunction with a switch, cartridge. The breech handle which carries part
squib, or solenoid, and the necessary electrical of the electrical circuitry is designed so that
circuitry. The governor system may be mechani- the breech assembly has to be closed and locked
cally connected to a bypass valve in the hot-gas before the breech electrical circuit can be com-
circuit. pleted. This is to prevent any inadvertent firing
Method 2. Fig.19 shows the arrangement of of the cartridge before the breech is completely
the cutout device which diverts the gas from com- closed and locked.
pletely going through the turbine wheel downstream
of the cartridge nozzle system. This approach to H Overpressure Protection
speed control has a sensing arrangement which may Overpressure protection is a significant de-
either sense the output speed, the output accel- sign requirement for the safe handling and utili-
eration, or the output torque of the starter. zation of the cartridge gas in a starter system.
There have been two schemes of control mechanisms There are two basic approaches to overpressure
with this type of design approach. The first protection. One design is to use a rupture dia-
would be to place some type of baffle plate be- phragm. The second is to use a spring-loaded
tween the nozzle and the turbine, so that the gas plunger. Both systems have been used satisfac-
coming from the nozzle would be deflected and pre- torily.
vented from going through the turbine. The second
would be to vary the position of the turbine with 4 SUMMARY
respect to the nozzles, such that the blades could
be moved out of the nozzle gas flow. In summary, the design of the cartridge-
Method 3. Fig.20 is a diagram showing two pneumatic starter begins when the prime contractor
approaches to aerodynamic or inherent turbine combines the engine starting requirements with the
speed control. The first arrangement is that of accessory loading and aircraft performance re-
using a torque-absorbing fan as a part of the ro- quirements to establish the starter performance
tating single-stage turbine assembly. The second limits.
approach shown is that of staging the turbine The starter manufacturer first analyzes the
wheel, thus producing a torque curve which in it- airframe requirements and approximates the starter
self goes to zero torque within a safe turbine output torque versus speed curve which will sat-
yield speed. There are some basic efficiency isfy a particular application. The size and type
losses in both systems; however, in the fan ap- of gas generator is determined to supply the re-
proach, the losses in the critical starter operat- quired rate of gas flow (namely, Wf, C o , and so
ing speeds can be kept at a minimum. forth) and the total propellant burn time. The

13
starter nozzles and turbine rotor are designed to In the early 1950's the U.S. Air Force haa
convert the gas energy to mechanical energy within its first jet engine cartridge starter operational
an economical weight and package size. A compro- on the B-57 aircraft. Beginning in the late
mise is made in the nozzle and turbine to provide 1950 1 s, cartridge-pneumatic starters were devel-
both cartridge and pneumatic-mode operation. oped and have been installed on operational F-100,
Special design considerations unique to the F-101, F-105, B-52, KC/C-135, F-4C, and F-111A
cartridge starter are given to the handling of the aircraft. It is evident from this lengthy list of
high-temperature and pressure cartridge gas to U.S. Air Force applications that the cartridge-
arrive at reasonable endurance life. Safety fea- pneumatic starter has proven itself to be a relia-
tures must be included to provide safe no-load ble means of providing self-sufficient jet engine
operation and overpressure relief. starting.

Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1967/79887/V001T01A048/2389775/v001t01a048-67-gt-49.pdf by guest on 06 October 2022

14