t~C-"~a,L ~............

- . ~ . ' . v - r , y , ~ '-~ ¢'.:,

<"~"I A B L ' S N t ' # f ( ! ' 4 T R. & M. N o . 3 1 4 3
BEDFORD. (zs, aoo)
A~.C. Technical Report

MINISTRY OF A V I A T I O N

AERONAUTICAL RESEARCH COUNCIL

REPORTS AND MEMORANDA

The Effect of Changes in the Stability

Derivatives on the Dynamic Behaviour
of a Torpedo

j. D. LAMBERT,
A d m i r a l t y Research L a b o r a t o r y

© Crowncopyrightx96o

LONDON : HER MAJESTY'S STATIONERY OFFICE

I96o
PRic~ 8s. 6 d . N~T
The Effect of Changes in the Stability Derivatives Oil
t h e Dynamic Behaviour of a Torpedo
By
j. D. LAMBERT,
Admiralty Research Laboratory

Reports and Memoranda No. 3 ~43"

March, 1956

Summary.--T~s report investigates the extent to which the dynamic behaviour of a torpedo is sensitive to changes
in its stability derivatives. The main object in carrying out the investigation was to provide guidance on the accuracy
of measurement of the stability derivatives that should be necessary for any given torpedo. The considerations of the
report are, however, also pertinent to the problem of deciding the effectiveness of possible changes in the design of a
torpedo, the dynamic behaviour of which is unsatisfactory. Illustrative examples are worked out in detail. The report
emphasises the importance of the so-called margin of stability.

1. Introduction.--The purpose of this report is to investigate t h e extent to which the dynamic

behaviour of a torpedo is sensitive to changes in its stability derivatives. Since dynamic
behaviour covers the whole class of possible motions of a torpedo, attention has had to be confined
to certain well defined aspects. The main object in carrying out the investigation was to provide
guidance on the accuracy of measurement of the stability derivatives that would be necessary
for any given torpedo : specifically, what error in predicted performance will given errors in
the stability derivatives cause ? The considerations of the report are, however, also pertinent
to the problem of deciding the effectiveness of possible changes in the design of a torpedo, the
dynamic behaviour of which is unsatisfactory.

2. The Motion of the T o r p e d o . - - W e consider motion in a vertical plane only, and neglect
buoyancy and trim effects. The treatment applies equally to •motion in a horizontal plane only.
The relevant equations of motion are
Z~¢ + Zq$ + Z~e~~ = m ~ V ( ~ - mlVO , 0 m
(a)
M ~ + M~O + M @ ~ = J y O , .. .. • g D I (2)
where
V z
speed of torpedo, assumed constant
angle of attack
0 = pitch angle
elevator angle
q = 6 ---- pitching rate
Z denotes the coefficient of a force normal to the torpedo axis
* Admiralty Research Laboratory Report ARL/R3/HY/13/0.
 M denotes the coefficient of a moment about the transverse horizontal axis
through the torpedo c.g.
Z~ OZ/~c~, etc.
m s total transverse mass of torpedo -"- m + K~m:
---- total longitudinal mass of torpedo ~ m + K~m:
= mass of torpedo-
~44f = mass of displaced fluid
Jy = total moment of inertia about the transverse horizontal axis through the
c.g. -"- Iy + K'Iy:
Iy z moment of inertia of torpedo about the transverse horizontal axis through
the c.g.
.[y: --= moment of inertia of displaced fluid about the transverse horizontal axis
through the c.g.
K', K1, Ks are Lamb's inertia coefficients for an equivalent ellipsoid.
The positive senses of the various parameters are illustrated in Fig. 1.
If we multiply each term of equations (1) and (2) by e -p~ and integrate with respect to the
time t between 0 and ~ throughout (denoting Laplace-transformed quantities by a bar) and
eliminate O we have
[m2VJ,fl ~ -- (JyZ~ + m2VM~)p + M~Z~ -- M~(mlV + Zq)]a
= [ l . Z , 2 + M,( IV + G ) -- M J , 3 : , . . . . . . . (3)
If we had found, instead, the equation connecting G or p0 with 3,, the left-hand side would have
been identical with that of equation (3). We write this left-hand side as

where [Alfis q- Asp + A3]~, )

A1 = m2VJy
As = -- J,z

A , = + MqZ~ -- M~(m~V + Zq)

-

It follows from equation (3) that the transient part of the solution for ~ (t) will be
J . . . . . . . . . . . (4)

21 e",' + ;~s e 's*" . . . . . . . . . . . . . . . . . (5)

where/~, and/zs, the decay constants of the motion are the roots of
AI~ s + A~/~ + A~ = 0 . . . . . . . . . . . . . . . . . . (6)
and 21 and ~. are constants.
In particular, if the elevators are locked at zero, the right-hand side of equation (3) disappears,
and the expression (5) represents the complete solution for the angle of attack ~, following a
disturbance.
A torpedo is said to have dynamic stability, if, when disturbed from a straight-line path, it
will again settle down to a straight-line path (but not necessarily the original straight-line path),
t h a t is, it tends to reduce its angle of attack to zero. If a dynamically unstable torpedo is
disturbed from its straight-line path, it will circle with smaller and smaller radius until the
linear analysLs used here _no longer applies. It is clear from equation (5) that the necessary and
2
sufficient condition for the torpedo to have dynamic stability is that the roots of equation (6)
have negative real parts. The necessary and sufficient condition for this is that A t , As and A3
all have the same sign"
A~ = m~VJy > 0
A s = - - JyZ~ - m~VMq > 0
since Z~ < 0, M~ < 0 for all conventional torpedoes. The criterion for dynamic stability is
therefore that Aa > 0. Since Z~Me > 0, we can write
G= 1 - -
M~(m~V + Zq) > 0 for dynamic stability . . . . .
Z~Mq (7)
G is called the margin of stability. The following Table indicates torpedo behaviour for different
values of G.

Stability G Controllability Application

Dynamically unstable <0
Marginally stable 0 Requires special control equipment No known application.

Dynamically stable 0.1

0.2
0.3
0.4
0.5
t Turns rapidly with small rudders ; hard to
control and maintain in straight flight.

Turns rapidIy with medium-sized rudders;

Homing torpedoes.

Homing torpedoes and

0.6 controls moderately well. straight-running torpedoes.
0.7 Turns rapidly with large rudders ; controls Straight-running torpedoes.
0.8 easily.
0.9
1.0
>1.0 l Requires very large rudders ; controls very
easily.
Straight-running torpedoes.

2.1. Circling M o t i o n . - - S u p p o s e the torpedo is moving steadily in a vertical circle of constant

radius R, with the following (constant) values of its parameters
q=O =0'', ~=~*', ~ d~*
~=0=0.

Putting these values in equations (1) and (2) and solving for 0* and ~* we have
O* M~Z~ e - - Z~M~
G ~, = Z~Mq . . . . . . . . . . . . . . (8)

G ~-~. = Z~M~ . . . . . . . . . . (9)

(We note that, since tile right-hand side of equations (8) and (9) are both negative for all con-
ventional torpedoes,
0* 6"
sgn ~ ---=sgn~fi, = -- sgn G.
This implies that a dynamically stable torpedo (G > 0) turns with its elevators, while a d y n a m i c -
ally unstable one (G < 0) turns against its elevators.)
3
 I n a stable (G > 0) turn of constant radius R, V = RO*, and from equation (8) we h a v e

R = VGZ~Mq 1
M~Z~ - - Z~M~ ~ ~* " ". . . . . . . . . . . . . (10)

3. The Effect of Errors in the Stability D e r i v a t i v e s . - - W e can n o w s t u d y t h e effects of errors in

t h e stability derivatives Z~, M~, Z~, M~,, Zq and M e on three aspects of the d y n a m i c b e h a v i o u r
of a torpedo :
(a) The effect on the radius of t u r n R for a given elevator angle ~,*
(b) The effect on the margin of stability G
(c) The effect on the transient m o t i o n of t h e torpedo following a disturbance. This is done
b y studying t h e effect on the decay constants ~z and ¢2 defined b y equation (5).

Errors in t h e range ± 20 per cent will be considered for the static and control surface
derivatives Z~, M~, Z~, and Me0, and errors in the range -¢- 50 per cent for the rotary derivatives
Z e and M e. E a c h case will be illustrated b y examples of two torpedoes of widely differing
h y d r o d y n a m i c characteristics, Torpedo A (G about 1.0), and Torpedo 13 (G about 0.6). T h e y
h a v e t h e following h y d r o d y n a m i c coefficients :

T O R P E D O A.
OCL 3" 09 " bCL aCt
' 08. -- 0.70 ; ~(l/R) -- -- 1.40~

OCM 0.055 ; OCM _ 0.37 ; ~C~

0~, O(IlR) - - -- 0"63..
We use the .relations
aCL
Z= = ½pA V ~ OC__~L' Ze . = ½oA V 2 ~CL~.o ," Ze = ½pA V1 0 (l/R)

Mo= ~pAV~I ~C~'~ , M~ = ~pAV~I ~C~'~ , M ~ = ½ p A V Z ~ ( IOCM

/R)

where p = density of water = 2 slugs/cu ft

A = m a x i m u m cross-sectionM area of torpedo = 2 . 4 ft ~

V = speed of torpedo = 40 ft/sec

l = length of torpedo = 14 ft.

This gives
Z~ Z,~e Zq
10~ - - 11"866; 103--2.688; 103-- 1"888
M~ Me e Me
103 -- 2"957 ; 103 -- 19"891 ; 103- 11"967.

Also mass of torpedo = 58.5 slugs

ary = m o m e n t of inertia of torpedo about the transverse horizontal axis t h r o u g h

the c.g. = 745 slugs/ft ~.
4
T h e L a m b inertia coefficients for an ellipsoid of t h e s a m e fineness ratio (8) are
K1 = 0" 029 ; K2 = 0" 945 ; K' = 0" 840, giving

m~V
10 a -
+ 2" 408 " m~V
10 a - + 4"551 ,"
Jy
• , 10 ~ - - q- 1"371 .

T O R P E D O B.
OCL ~CL ~CL
8e
_ _

2-29,
. _

~8o
_ - -

0.396,
o

~(I/R) _ _
1.04

aCM ~CM ~CM

80~ b 0.556, ~, 0-229 • ~(I/R) 0.50

p - - 2 slugs/cu It m = 94.47 slugs

A = 2"405 fff Iy = 1886" 8 slugs/It 2

V = 49 ft/sec Fineness ratio - - 11.7, w h e n c e

l = 2 0 . 4 9 It K1 = 0 . 0 1 9 ; K~ = 0.968 ; K' = 0.908.

These give
Z2 Z~ Z~
103 - - 13.223 ; 10 ~ - - 2.287 ; 103 - - 2-511

M~ M~ M~
103-+65"785; 103 = - -
g

27" 0 9 5 ;
_

10 ~
_ - -
24. 738

V - V j,
10 a -- +9"110" 10 a ---¢-4"717" -- b 3-600.
, , 10 ~

3.1. The Effect on Radius of T u r n . - - F o r a given elevator deflection dr*, the radius pf t u r n is
(equation (10)).
R=R' V " }
dr*
where . . . . . . . (11)

R' = 7 - 77V
W e denote b y R0 a n d Ro' t h e values of R a n d R' w h e n t h e r e are no errors in the stability
derivatives, a n d b y ~R a n d OR' t h e changes in R a n d R ' due to changes aC in C, where C is
one of Z~, M~, Zo , M~,, Zq a n d Mq.

Since V a n d d,* are constant, it is clear t h a t

~R ~R'
R R'"
T h e fractional change in R for a n y given fractional error in C can be calculated from e q u a t i o n (11)
as set d o w n below, for alt six i n t e r p r e t a t i o n s of C.

W e n o t e t h a t Ro has t h e following v a l u e s for the two torpedoes chosen as examples •

T o r p e d o A Ro = 144 ft w h e n ~ * = 10 deg
T o r p e d o B R0 = 100 It w h e n dd" = 10 deg.
5
( 7 8 2 9 0 ) A *
Errors in Z~ Z~-+Z~+OZ~
aR M~EMqZ% -- M%(m~V + Zq)]
R MqZ~, - - M~,(mfl7 + Zq) 2,5 e
M~Z~
~R aZ~/Z~
Torpedo A •
R oZ~
-- 21.95 -- 22.71 Z~

~R
Torpedo 13 •
R oZ~"
+ 0 . 9 2 + 0 . 6 4 Z~

Errors in M~ M~-~M~+OM~
~M~IM~
R MqZ~ -- M~(m~V + Zq) md e
Z~M~ (1 +-#7/
Torpedo A •
R dM~
+ 21.95 - - 0"77 M=

Torpedo B •
~R ~M~/M~ .
R ~M~ "
--0.92--0.27 M~

Errors in Z%
~R dZ~JZ~
R - - M% Z~ 1 -- - -
M~ Z~ Z~
~R ~Z~ /Z~
Torpedo A •
28.70 dZ ~
Z%
~R dZo /Z~,
Torpedo B •
R
-- 3" 38 ~Z~
Z~
Errors in Mo~ M6~ --> M,~ + ~M~
~R
R
1
Z~ M~ Mot
~R
Torpedo A •
R
- - O"97 0M~
mo¢
8R
Torpedo B "
R (gm~ e
1.42
M~
- -
Errors in Zq Z~ ~ Zq + (~Zq
~R --"Mo:Z q ~Zq __ Zq ~0 - - 1 (}Zf[
R - - ~/iqZo: - - ]V_fa(~/]~Ig + Zq) Zq (]/y/,lV+ Zq) G O Zq '
where Go is the value oi G, the margin of stability, when there are no errors in the stability
derivatives.

Torpedo A : , R __ 0"04 dZq

R Zq
dR ~Zq
Torpedo B : R --+0"91 Zq "

Errors in M e Mq"-+ Mq + 6Mq

~R MqZ~ ~ M q _ 1 ~M~
R -- -- M (mlV + &) Go Mg

Torpedo A • ~R
R - +0.99 ~Mq

bR cSMq
Torpedo B " R -- + 1"80 M-~-"

These results are plotted in the form percentage error in R against percentage error in C in Fig. 2
for Torpedo A, and in Fig. 3 for Torpedo ]3. It is clear from Fig. 2 that, for Torpedo A, errors
only in Mq and M~, are significant. It is therefore useful to study the variation of R when there
are errors in Mq and M~ simultaneously. The result for Torpedo A is
6Mq 6M,~~
+ 0 " 9 9 - - -- 1 " 0 4 - -
~R Mq M~e
R ~M,~
1 + 1.04--
M~
T h i s c a n be plotted as a family of straight lines in the ~R/R -- bMq/Mq plane with ~M~,/M~ as
parameter. From this it Call be seen what ranges of errors (positive and negative) in M~ and
M~ are permissible for a given permissible range of error in R. This information is plotted in
Fig. 4.

For Torpedo B errors in all stability derivatives are: significant, and there is no point in
considering simultaneous variations of two only.

3.2. The Effect on the Margin of Stability.--G was defined by equation (7) as

GO --- 1 - - M ~ ( m l Y + Zq)
Z~Mq
where Go is the value of G when there are no errors in the stability derivatives. We are now
interested in the value of G when errors in the stability derivatives exist, and not in the fractional
change in G. The values of Go for the two torpedoes being considered are
Torpedo A : Go = + 1.011
Torpedo B : Go = + 0. 556.
7
(78290) A*2
Errors in Z~ Z~-~ Z ~ + o Z ~
Go-- 1
G=I+
OZ~
l+z~
0.011
Torpedo A : G=I+- OZ~
1 + Z~
0.444
Torpedo B : G=I--
~Z~
1+ z~
Errors in M , M ~ - ~ M ~ + OM~

G=Go-F(Go--1)~M~
M~
~M~
Torpedo A • G = 1.011 + 0-011 M~

Torpedo B • G ---- 0.556 -- 0. 444 aM~

M~
Errors in Zq Zq-->" Zq ~-- aZq

G = Go + mlV + Zq (Go- 1) Z
OZq
Torpedo A • G = 1.011 -- 0" 040

" OZq
Torpedo B " G ---- 0-556 + 0" 050 Zq "

Errors in Mq Mq --+ Mq + (~Mq

Go -- 1
G=I+
aMq
1+
0.011
Torpedo A : G = 1+
~Mg
1 +Mq

0.444
Torpedo B • G = I aMq which is the s a m e variation as for OZ~
1+ Mq' Z~

These results are plotted with OC/C as a percentage in Fig. 5 for Torpedo A and in Fig. 6 for
Torpedo B. I t is obvious from the form of the equations t h a t the variation of G w i t h errors in
t h e derivatives decreases as Go approaches u n i t y and is in fact zero at Go = 1.

3.3. The Effect on the Transient Motion of the TorTbedo, Following a Disturbance.--It was shown
in Section 2 t h a t the t r a n s i e n t p a r t of the solution for t h e angle of a t t a c k ~(t) following a
d i s t u r b a n c e was the expression (5) :
~1 e ~'1~ + 2~ e~ ~ .
8
 The transient solution for the depth Zo, or pitching rate 0 would be of the same form, with of
course, different values of the constants 2~ and ~2. Real values of ~ and ~ will be associated
with aperiodic motion, and imaginary values with oscillatory motion.
The effect of errors in the stability derivatives on the transient motion of tile torpedo can be
studied in two sub-sections :
(a) The effect of such errors on the decay constants ~ and ~2
(b) The effect of such errors on the transient motion following one particular disturbance
which will be taken as a step function input to the elevators.

3.3 (a).--The effect of errors on the decay constants.--The decay constants were defined b y
equations (4) and (6). It is obvious from these that there are two types of problem involved
since errors in Zq or M~ cause A8 only to vary, while errors in Z~ or Me cause both A2 and A3 to
vary.
Errors in M~ M~ --> M~ + ~M~
Let # be a root of the new equation (replacing equation (6))

Al:d + A2~ + As -- (miV + Zq)M~ ~M~

M~ - 0 .
Put # ----y and ~M~/M~ = x, and this becomes the equation of a conic in the x - y plane. In
conventional conic notation, it becomes
b~y~ + 2g~x + 2fly + cl --- O,
where bl = + A1 = m2VJy
2g~ = -- M~(m~V + Zq)
2fl = + A2 = -- m~VMq --J,Z~,
cl = + As = MqZ~ -- m~(m~V + Zq).
The discriminant A is, in conic notation, h~~ -- a~b~ = 0. Hence the equation above represents
a parabola, providing the conic is non-degenerate (the case where the conic is degenerate is
discussed below). The parabola passes through the points (0, ~ ) and (0, ~ ) and its axis is parallel
to the x axis. Its vertex has an x co-ordinate of
' A - _ 1 - ( 2VM -- J, zo)
2b~g~ 4m~VJy M~,(m~V + Ze) "
The value of the decay constants for any given error in M~, say OM~*, are the values of y at
which the line x = ~M~*/M¢ meets the parabola.
The parabola cuts the x axis at the point x = Go~(1 -- Go), y = 0, where Go is the margin of
stability calculated when no errors exist in any derivative. With this value of x, t h e torpedo is
marginally dynamically stable. Moreover, the nearer Go is to unity the smaller is the change in
the decay constants for any given error. At Go = 1, the coefficient gl in the equation of the
parabola disappears, and this is t h e condition for the parabola to degenerate into a parallel
line-pa!r in the direction of the x axis, which implies no change at all in the decay constants for
errors m M~. W e assume that when no errors exist, the torpedo is dynamically stable, t h a t is
Go > 0 and ~z and s~ negative. It follows that the parabola faces right or left according as
G0~l.
The parabola is plotted in Fig. 7 for Torpedo A, and in Fig. 8 for Torpedo B. It should be
noticed that the horizontal scales of these diagrams are in units of ~M~/M~ and not (~M~/M~)
per cent as in previous diagrams. The variations of the decay constants are greater for
9
Torpedo B than for Torpedo A, as is to be expected, since Go is nearer unity for Torpedo A. In
dM~
fact, for Torpedo A, over the range ~ ~< 0.2 (ile., ~ 20 per cent error), there is no noticeable
change in the decay constants. For Torpedo B the change in the decay constants for the same
range of ~M~/M~ is noticeable but no t significant•
The torpedo is dynamically stable or unstable according as #1 and ~2 have negative or positive
real parts• When ~1 and t*2 become imaginary (i.e., in the region of the diagram past the vertex
of the parabola), the motion hitherto aperiodic becomes oscillatory. That the oscillatory motion
is, in fact, stable can be easily checked.
Errors in Zq Zq-+ Zq 4- ~Zq
Let # be a root of the new equation
Axl~2 Jr- A2¢ + A3 -- M~Zqc}Zq
Zq -- O.
Put/z = y and OZq/Zq = x and we can write this in conic notation as before
b~y e 4- 2g~x 4- 2fey + ee = 0 ,
where be = + A1 = meVJy
2ge = -- M~Zq
2re = + A e = -- m2VMq -- J,Z~
ce = + Aa = MqZ~ -- M~(m~V + Zq).
This is, again, a parabola passing through (0, /~) and (0, ~e). The x co-ordinate of the vertex
is now
A e - b2ce I : (meVMq - - J , Z ~ ) 2_ ] m~V + Z e
2beg~ ~-- 1 -- 4meVJyM~(m~V + Ze)J Ze
It will meet the x axis where
X-- Go m~V + Z e
1 - - Go Zq
It is in fact the same parabola as before, but with the horizontal scale multiplied b y a factor
(m~V + Ze)/Z e. Minimum variation again occurs when Go = 1, when the parabola degenerates
as before. The parabola is plotted in Fig. 7 for Torpedo A and Fig. 8 for Torpedo B. In both
cases the variation of the decay constants is a little greater than for the M~ case but it is still
negligible for Torpedo A and not very significant for Torpedo ]3 in the range _~[e < 0" 2.
* /

Errors in M e M e --~ M e + ~Mq

L e t / , be a root of the new equation
/ ~Me~ ~Me
A ~ e + ~A2 - - m e V M e -M~q ] /* + A8 + Z~M~ Me - - 0 P

P u t y = / z and x = dMq/Mq. In conic notation the equation becomes

2h3xy 4- bay e 4- 2&x 4- 2fay + ca = 0 . . . . . . . . . (12)
where
2ha = -- meVMq.

293 = + Z~Mq (13)

2f~ = + A~ = - - J ~ & - m~VM~
ca = + Aa = M J ~ -- M ~ ( m l V + Zq) l
10
The discriminant A = h~~ - - a~b~ = hd > O, so the equation represents a conic which, if non-
degenerate, is a hyperbola. (The case when the conic is degenerate.will be discussed below). The
equation of the asymptotes is got from this equation by adding a constant z such that
h3 g~
ha b~ f~ = 0.
g~ f~ c~ +
Solving for ~ we get
c3 + ~ = 2& ( 4Ah~ --22bsg3~j = 2 g 3 ,

since 4f3h8 - - 2b~g~ = 4h~ ~ from (13).

The asymptote pair has, therefore, the equation
2h~xy + b~y ~ + 2 & x + 2f~y + 2& = 0 . . . . . . . . . . (14)
The absence of a term in x ~ shows that one of the asymptotes is parallel to the x axis. The slope
of t h e other one is therefore the tangent of the angle between them and is
2~¢/(ha ~ - - a a b a ) Mq
+- + -- J7 "
From (14) we see that the point (-- 1, 0) lies on the asymptote pair, and since the horizontal
asymptote is certainly not y = 0, the point (-- 1, 0) necessarily lies on the sloping asymptote,
whose equation is therefore
Mq (1 + x)

Since the hyperbola passes through the points (0, /~1) and (0, /~2) where/~1 and #2 are negative,
this asymptote must have a negative gradient, whence its equation is
Mq (1 -¢- x)

Mq being negative for all conventional torpedoes. The equation of the other asymptote is found
by differentiating equation (14) and finding the value of y for which d y / d x vanishes. It is
g~ Z~
Y -- h8 - - m 2 V "

The horizontal asymptote has therefore the equation

Z~
Y = m~V "
We note that the asymptotes intersect at (x*, y*), where
x* = JyZ~ - - m 2 V M q
m=VMq
We can now draw the asymptotes directly, and_ we know, moreover, two points on the hyperbola,
namely, (0,/~) and (0,/,~). There is o n e other point of interest on the hyperbola. From
equation (12) the x axis cuts the hyperbola where
-- 68
x-- -- Go.

There are four possible configurations of the hyperbola depending on whether x* X 0 and
Go <> 1. These are shown in Fig. 9. If we use the fact that the intercepts on any straight line
cut off between a hyperbola and its asymptotes are equal, it is possible to sketch in the hyperbola
11
with reasonable accuracy from a knowledge of its asymptotes, the points (0,/,,), (0, ~ ) and
(-- Go, 0), which are known to lie on it. In the case Go < 1, as Go --> 1, the rate of variation of
one decay constant decreases, while that of the other increases to the slope of the sloping
asymptote. In the case Go > 1 it is clear that the variation of both decay constants decreases
as Go approaches unity. If Go = 1, the hyperbola degenerates into its asymptotes, and only one
decay constant varies.

The hyperbola for Torpedo A is shown in Fig. 10, and for Torpedo B in Fig. 11, and the
.stabilit.y regions are shown for each. It is easily proved that the region of oscillatory motion
is a region of stable motion. It is interesting to note that when Go < 1 it is impossible to reach
a condition of oscillatory motion of the body by altering M e only.

It is clear from these Figures that errors in M e are far more significant as regards the decay
constants, than are errors in Z~ and M~. In fact an error of -- 60 per cent in M~ would cause
Torpedo A to oscillate, and Tori~edo B to become dynamically unstable.

Errors in Z~ Z~.--> Z~ q- $Z~

L e t / , be a root of the new equation

A I~ ~ +
(A ~ - - ~z4
J , . & 72~1 ~* + A 3 + M J .
~z~
& _ O.

Putting/~ = y and dZ~/Z~. = X, this equation becomes, in conic notation.

2h4xy + b~y ~ + 2g,x + 2f4y + ca = O,

where
- -J,&
b~ = + A1 = m,~vJ,
2g,=+ MqZ~

2A=+ A2 - - JyZ~ - m2VM e

C 4 ~ -{- A3 = MeZ~ - - M~(r~lV + Z~).

This is again a hyperbola, and, in the same way as before, we find that the asymptotes have the
equations
y = + - - Me
jy (horizontal asymptote)

y= + ~ Z~ (1 + x) (sloping asymptote) .

They intersect in the point (x*, y*) where

x* = m2VMq - - JyZ~

Since the x axis cuts the hyperbola at x = -- c4/2g~ =- - - Go, as before, the remarks made about
the significance of having a value of Go close to unity still apply. The four configurations shown
in Fig. 9 also apply, if the new expression for x* is used. These hyperbolae are'plotted in Fig. 10
for Torpedo A and in Fig. 11 for Torpedo B. The variations in the decay constants are still
large, but not so.seriously as they were for errors in Mq, particularly as regards measuring accuracy,
since accuracy m measuring Z~ is far easier to achieve than accuracy in measuring M v For
Torpedo A an error of -- 100 per cent would be necessary to cause instability and an error of
+ 200 per cent to cause oscillatory motion. For Torpedo 13, instability would occur when Z~
12
had an error of - 60 per cent. For both torpedoes, x* is positive for the Z~ case and negative
-

for the M s case. This implies t h a t the sloping asymptote has a less steep gradient in the Z~ case
than in the M s case, and that the variations in the decay constants are correspondingly less.

3.3• (b). The elect Of errors on the transient motion for one particular disturbance.--The
disturbance will be t a k e n as a step function input on the elevators. The subsequent solution for
the angle of attack will be studied. The relevant equation is equation (3), where ~(t) is now a
step function of magnitude ~,*. Then,
1
<(p) = }

and equation (3) gives,

~(p) ],z~,p + M~o(my + z~) - MsZ~,
<* m W J , p ( p - #I)(P - #3)

b y the definition of #i and #3. Splitting the right-hand side into partial fractions we have

a(p)_h=q h~q h2 . . . . . . . . . . . (is)

where
h = M o b ( m y + Zs) -- MsZ~ ~
m2VJy #i#2

hi = ]yZ~#l + M%(mlV + Zs) -- M J % (16)

m.VJ~#d#l- #2)
h~ = J,Z~.#~ -}- M%(mlV -J- Zs) -- MqZo.

Inverse Laplace-transforming equation (15) gives

~~*
( t ) = ha + hle~U + ~2 e'*2' .

Since we are interested only in the transient solution, and not in the steady-state solution (which
is ha), we divide b y ~3 to get finally',

~(t) _ 1 + hI e ~ l ' + hl e.2'

where
hl- hi F /,z,,#1 ] (17)
h~ - I_M~JmY + Zs) -- MsZ~ + 1 #1 -- #2

hi-L-- M~ (m~V + Zs) -- MJoo + I - - #3 -- #1

#i and #3 are affected b y errors in Z~, M=, Zs and M s as already shown, hl and ,~1 are affected
by errors in all six derivatives. It is therefore possible to study how the solution (16) varies
with errors in each of the six stability derivatives, one at a time. This has been done for three
13
values of error in each derivative, namely 0, ~ 50 per cent for the rotary derivatives Zq and Mq,
and 0, ± 20 per cent for the others. The results for Torpedo A are containedin Fig. 12, and for
Torpedo B in Fig. 13. The time for the ordinate to reach 95 per cent of its final value is marked
in each case. Errors in Z~c and M~e do not affect either torpedo noticeably. For the remaining
derivatives, errors appear to affect Torpedo B more adversely than they do Torpedo A particularly
in the case of the rotary derivatives Z e and Mq. An error of - - 5 0 per cent in M e causes a
substantial change in tile motion of Torpedo B. It should-be noticed t h a t the time to reach
95 per cent of the final value is less for Torpedo A than for Torpedo B ; this is to be expected
since Torpedo A has a larger margin of stability.

4. Summary a~cd Comlusiom.--In this report, the extent to which the dynamic behaviour of
the torpedo is sensitive to changes in its stability derivatives has been investigated. Attention
has necessarily been confined to certain well defined aspects of dynamic behaviour. These aspects
were the radius of turn for a given elevator angle, the margin of stability, the decay constants of
disturbed motion, and the motion following a particular disturbance, namely, a step function
input to tile elevators. It is not too unreasonable to suppose t h a t these aspects are broadly
representative of dynamic behaviour. It must be admitted, however, that the theoretical results
apply to an uncontrolled torpedo. Nevertheless, it should be noted that according to tile Table,
the margin of stability indicates the ease with which a control system for a homing torpedo can
be designed.
The results obtained in particular cases, namely, Torpedo A and Torpedo B which have been
used as illustrative examples, m a y be summarised as follows : The radius of turn per elevator
angle of Torpedo A is very susceptible to errors in M~c and M e ; t h a t of Torpedo 13 is very
susceptible to errors in all derivatives except perhaps Zo. The margin of stability G, for Torpedo
A varies very little with errors in the stability derivatives. For Torpedo B, G varies rapidly
with errors in Z~; Mq and M~. For both torpedoes, tile decay constants v a r y much more with
errors i n Z~ and M e than with errors in M~ and Zq. This tendency is reflected in the effect of
errors on the solution for angle of attack following a step function input to the elevators, but it
is not as pronounced as one would expect, presumably due to the effects of the errors on tile
coefficients ~1 and ~1. For Torpedo A, the variation of the solution is small for all feasible errors.
This is not so for Torpedo B, the variations due to errors in Z~ and Mq being rather severe.
In view of tile complexity of the concept of dynamic behaviour and tile number of parameters
involved, it is difficult to draw general conclusions. It does seem clear, however, that the
susceptibility of torpedo performance to changes or errors in tile stability derivatives depends
to a great extent on the margin of stability. Tile effect of errors is, in most respects, at a
minimum when Go = 1, that is, when the torpedo is marginally statically stable.

A cknowledgements.--The author is indebted to Mr. I. J. Campbell of the Admiralty Research

Laboratory and to Mr. A. MacDonald of the Torpedo Experimental Establishment, for much
helpful advice given during the preparation of this report.

14
 ' ' @ ~ V

~.
FIG. 1. Sign convention in pitch plane.
FIo. 2. Percentage error in ;adius of turn R against percentage error in hydro:
dynamic coefficients C Torpedo A).

16
 \ /
/C= Mt

,\ 50 I
¢
1, i
\ 4o, ' I
! /

C --Z-c

1 ,
o

:ol I

.50 "40
I
"v,~o +'qO + 50
"lO "li¢~~
f,
C=Z~ e

/ \
\
/
l
¢ t

\
'l
/

FIG. 3. Percentage error in radius of turn R against percentage error in hydro-

dynamic coefficients C (Torpedo A). . . . . . . . .

17
 zo?,

~o9,

"~ ~1,,o, \ \\ \ \\
J0~, 20% ~o~ 4o~ soT, ~o~

FIG. 4. Ranges for errors in M~ e and Mq for various

permissible errors in R (Torpedo A).

I' IC

t'Olt

1.0~

I'0X t ' = Mo¢

'~.___.____~ C = Z~,Mq
t,o ---.-=---- .__._=_____
C=7.~ L

0'9~

O,$

0.~0

-50~ -4.0% "50% -2og -1o%

O.tt5
o%
l
+ZO% ÷20% 4.5096 +,¢o~

FIG. 5. The variation of G, the margin of stability with errors in the stability
coefficients (Torpedo A).

18
 M~

FIG. 6. The variation of G, the margin of stability, with errors in the stability
derivatives (Torpedo }3).

19
 ~4
I:P/NAMIC STABILITY DYNAMIC STABILITY DYNAMIC
~-- - - DAMPED OSCILLATORYMOTION-----~ ,:~" APERIODICMOTION ~i~ INSTABILITY
I
4"9 z,~ I

"20 4.10 4.1o +fo +To +t!,o +eo ~ = . .

-2

DYNAMIC
STA I~dI..ITY
.,,t ~':~t' I:P/NAMfC STASI LITY
APERIODIC MOTION
DYNAMIC.
~r,~-INSTA~ILITY ~-
OSCILLATORY I ( -~M= ~ I ('~M" ,,
MOTION
(: ~tIM.<
"~,,) I

FIG. 7. The variation of the decay constants/~1, ~'2, with 6ZqtZq, - - 6 M ~ I M , (Torpedo A).

I
I
I
I
I
DYNAMIC INSTABILITY +4- DYNAMI~, STA~ILITY'APERIODICMOTION __~,J..~_.DYNAM IC - -
STAbILITy
DAMPED 05CILLATOR.Y
+3 MOTION
4.Z
I
I(7:)
I
+l I I
~ . 1 , i, I

iI , -4` )', )
I l 1', /t
t I "~ , /jJ//" I
. . . . ', / / f
I I i ~ . !DY,,,AM,C
I I -7- ~ " I 5TA BILl T"/"
] ' I /~12~ ~ i ! DAMPED
I
=- ~ . ~ ~ I::)YNAMIC STABILIT'Y: APER[0OIC MoTiON ,...~5Cl LLATOI~Y

q" ' I I "'° I

DYNAMIC IMSTAI~ILtT7 •[ I -il. I
( - 6 M ~ / M ~.) 1
I I i
I I -J2, I
I
I I .~. I
i I

FIG. 8. The variation of the decay constants/~1,/~2, with errors in Z Gand M~ (Torpedo B).
20
 Go ~'1

(-Qo,O)~< •
(-I,0~ ,, X.

(ii)

"" ": 0

(~1)/

\ \
( iii ) (iv)

"FIG. 9.

21
 "1
I

I
-b4, -I.2 "'¢ ".Z "H.O H'Z.O *~.0. 'x

DYNAMIC
DYNAMIC
IN$TA~ILITY STABILITY
APERIODIC, MOTION

"~<1
E~
z~

.~zo, • l
X " Z---~ l
I I
DYNAMIC DYNAMIC STAI~ILITY IOYNA~CSTA~ILITYI
DYNAMIC
I N.~'r'AEli LIT "/ APERIODI~ MOTION ]DAMPEDOSCILLAToRY]~TA~ILITy,
( ~-~- ) I MoTiON ( S£Z~ ) AF'E~,oDJCm0T,oN
I

FIG. 10. Variation of decay constants/~1 and/~2 with ~Z~/Z=and aMq/Mq(Torpedo A).

*'1 +'2 .l¢ 4~ 4,,,B ¢1'0 , , , q30 +~'0 -t.,,,FO

x,. x • ~._MM~
MR,

X=J;Z,~
'Z~

DYNAMIC DYNAMIC BTA~I LITY

<=-'IN STA~ iLI T . ~ 7 - - - ~ APERIODIC MOTION

-12

= SM,t
-I¢ M,.

I -15
I

FIG. 11. Variation of decay constants/~1 and/*2 with $Z~/Z~ and ~M~/Mq(Torpedo ]3).

g2
 I.O

O.B
O'B ~ - =- ZO~

0,7
,,
O.7- 1
ERRORS OF ~ 2 0 ~ IN M X HAVE
i I NO EFFECT FOI~ THI~. TORPEDO.
O.E- 1

0,5 ..... 1
0"4- . . . . .
I I 0,~ - - 1

]I
II
0,5 .....
II I'I O.N - - 11

0,~ . . . . I I o'2 - - II
I II
I
O,I . - I
O'l ~ Jl
........
',I
I..~I .... I .... I .... I_ o ......... I,II ..I .... I ...........
0"5 I'0 1"5 2"0 2~ ~'0 o.~ I,O P5 Z.O '~'5 3'O
~ma ~, (SEES.)
(i) ERRORS IN Z~ (ii) ERRORS IN M,~ (ll0 ERRORS IN Zse
~(t) •4 (e') ~(~-)
c-W*
I,¢

_/....
O'8

O'7 M~¢ 0'7

i ,' sr~a = 0%
O.G Ill ~q, .

0"5 0,5
IIi I
II
II

II
II
O'Z II
i1
I
(PI Oq II
Ii
.[
Ii
,,=
0.5 ~'o I-5 I1~ ~,~ ~'O O'5 I,o P5 2"0 Z'5 3'0 o .... o ~ . . . . I,O t,5 ~'0 ~'5 3'O
Time C, ( 5 ~ c s ) T,ME I: (SEes.) T, M E =, (sEes.)
(iv) ERRORS IN M~e (v)'ERRORS IN Z,~ (vJ)ERRORS IN M,u

oC(~p..) = ANGLE OF A T T A C K FOLLOWING STEP' FLINC.TIOIXl INF'~T ON ELEVATORS.

¢~¢~" = MAGNITUDE OF THE STEP FLINC.TION.

C = STEADY STATE VALUE OF ANGLE oF ATTACK.

FIG. 12. The effect of errors on t o r p e d o m o t i o n for T o r p e d o A.

23
"G

I'0

/< /i
04
,,'/ o
~ ¢-" ~ I

c~
O,e /~'~t I Z. = +20% 0.t
k I I ,~z,~ = 0%

/I//e.,_l
o'7 i I Z,~. 0,;
E'RROR,50P -+ RO%. IN Z~¢ HAVE
i_
04 I I 2"= 0.:: I I I M,x. NO EFFECT FOR THI.~ "I"ORPSO0,~
I i
0".= I i o,s
?- / I i I II
i I
0,4 {>4-
I I t l
I 1 0,: I [ I
I I
O,Z I I II 0.~ I [
fl ' I I
i 1 I j~ I
0,1 i i o.i I I I
1 I ~ V I 1 I
0 J I 1 . . . . , I ,I .-
0 J.o z.o .~.o ~'.o 5.o eo ¢o s.o ~.o ~o~ O o Po ~,0 3"o 4"0 £,.o ~.o 7:o s.o s.o ~o.o

(i) ERRORS IN Z~ (ii) ERRORS 1N Me¢ (iii) ERRORS IN Z~¢

~..[1:)

I,O
1:'0
0"9
/ ',
0"8 I Me

0.7
ERROR..% OF" f- 2 0 ~ IN M f= HAV~
NO EFFECT ON THIS TORP~-I~O.
/ i L[ z~, " ~°gll o5
t [ l
04
I I I

0"3 ~/ • I I I
I I I I I/f
O"Z I
I
I
I 0.2
I I I I
0"! /----- I 1 I O . It - - I
I I I I
I . , I , I, 1 •
I.o ~o 3.0 4.0 5"~ ~,o r.'o s.o ~.o ,o-o o ,.o ~o ~,o ,o s'.o ¢:o % t;.o ~;.o ,o.o
T~M~ ~. (sEes) T,ME =, (s~cs.)
(iv) ERRORS IN M~,¢ (v) ERRORS IN liiZ~ (vi)ERRORS IN M a

e¢{'l:} = ANGLE OF ATTACK FOLLOWING STEP FUNCTION INPUT ON ELEVATORS.

~re~ = MAGNITUDE OF THE STEP FUNCTIOIM,
C = STEADY STATE VALUE OF ANQLE OP' A T T A C K .

FIG. 13. The effect of errors on torpedo motion for Torpedo ]3.
 R. & M. No. 3143

Publications of the
Aeronautical Research Council
A N N U A L T E C H N I C A L R E P O R T S OF THE A E R O N A U T I C A L
RESEARCH COUNCIL (BOUND VOLUMES)
1939 Vol. L Aerodynamics General, Performance, Airscrews, Engines. 5os. (52s.).
Vol. II. Stability and Control, Flutter and Vibration, Instruments, Structures, Seaplanes, etc.
63s. (65s.)
I94o Aero and Hydrodynamics, Aerofoils, Airscrews, Engines, Flutter, Icing, Stability and Control,
Structures, and a miseelhneous section. SOS. (52s.)
194I Aero and Hydrodynamics, Aerofoils, Airscrews, Engines, Flutter, Stability and Control,
Structures. 63s. (65s.)
1942 Vol. I. Aero and Hydrodynamics, Aerofoils, Airscrews, Engines. 75s. (77s.)
Vol. II. Noise, Parachutes, Stability and Control, Structures, Vibration, Wind Tunnels.
47s. 6d. (49 s. 6d.)
1943 Vol. I. Aerodynamics, Aerofoils, Airscrews. Sos. (82s.)
Vol. II. Engines, Flutter, Materials, Parachutes, Performance, Stability and Control, Structures.
9os. (92s. 9d.)
t944 Vol. I. Aero and Hydrodynamics, Aerofoils, Aircraft, Airscrews, Controls. 84s. (86s. 6d.)
Vol. IL Flutter and Vibration, Materials, Miscellaneous, Navigation, Parachutes, Performance,
Plates and Panels, Stability, Structures, Test Equipment, Wind Tunnels.
84s. (86s. 6d.)
I945 Vol. I. Aero and Hydrodynamics, Aerofoils. 13os. (I32S. 9d.)
Vol. II. Aircraft, Airscrews, Controls. I3OS. (I32S. 9d.)
Vol. III. Flutter and Vibration, Instruments, Miscellaneous, Parachutes, Plates and Panels,
PropuIsion. i3os. (i32s. 6,/.)
Vol. IV. Stability, Structures, Wind Tunnels, Wind Tunnel Technique. I SOS. (I 32s. 6d.)
Annual Reports of the Aeronautical Research C o u n c i l ~
I937 2s. (2s. 2d.) 1938 is. 6d. (IS. 8d.) 1939-48 3s. (3s. 5d.)
Index to all Reports and Memoranda published in the Annual
Technical Reports, and separately--
April, 195o - R. & M. 2600 2s. 6d. (zs. rod.)
Author Index to all Reports and Memoranda of the Aeronautical
Research Council--
19og--January, I954 R. & M. No. 2570 15s. (I5S. 8d.)
Indexes to the Technical Reports of the Aeronautical Research
Council--
December I, 1936--June 3o, I939 R. & M. No. 185o is. 3d. (IS. 5d.)
July I, I939--June 3o, I945 R. & M. No. 195o IS. (IS. 2d.)
July I, 1945--June 3o, I946 R. & M. No. 2o5o IS. (IS. 2d.)
july i, 1946---December 3r, x946 R. & M. No. 215o xs. 3~. (IS. 5d.)
January 1, 1947--June 30, 1947 R. & M. No. 225o IS. 3d. (IS. 5d.)

Published Reports and Memoranda of the Aeronautical Research

Council--
Between Nos. 2251-2349 R. & M. No. 235 ° is. 9 d. (is. ilg.)
Between Nos. 2351-2449 R. & M. No. 24.50 2s. (2s. 2d.)
Between Nos. 2451-2549 R. & M. No. 255 ° 2s. 6d. (2s. iod.)
Between Nos. 2551-2649 R. & M. No. 265o 2s. 6d. (2s. iod.)
Between Nos. 2651-2749 R. & M. No. 275o 2s. 6d. (2s. Iod.)
Prices in ~rackels include postage

HER MAJESTY'S STATIONERY OFFICE

York House, Kingsway, London W.C.z ; q.z3 Oxford Street, London W.I ~ t3a Castle Street, Edinburgh z~
39 King Street, Manchester 2 ~ z Edmund Street, Birmingham 3 ; t°9 St. Mary Street, Cardiff~ Tower Lane, Bristol I ;
80 Chiehester Street, Belfast,or through an3 bookseller.

S.O. Code No. 23-3143

R. & M. No. 314: