TECNJ URIT

r
Spm Labor

A
Mekeiweg Z 2b C eÎN
TeL O15.788813. FaIc O1S78183S

10. Füll-scale Measurement of Frequency Response

Characteristics of Ships in Steering
Parallel Shift Manoeuvre Test as Applied, to Actual Ships-
Keiichi KARASUNO*, Member

(Fro,n J.S.N.A. Japan, Vol: 128, Dcc. 1970)

Summary
This paper shows the resdits obtained on actual ships by using the method presented
before in A New Procedure of Manoeuvring Model Experiment i) The purpose of this
paper is to determine these four characteristic constants K' Ti' T2' and r3' of actual ships
by means of frequency response analysis of a parallel shift manoeuvring and trapezoidal
steering tests with fixed periods This procedure gives enough informations of frequency
response characteristics of a ship at practically important frequencies.
As an additional attempt this paper deals with a çb-çi phase plane analysis on a parallel
shift manoeüvre to determine the characteristic constants K', Ti', T2' and Ti'.
Full size experiments were carried out for eight actual ships ranging from a 360 GT
training ship to a 220,000 DWT oil tanker.
The conclusiòns obtained are that:
the combination of the parallel shift manoeuvre and periodic trapezoidal steering is
a promising procedure for full size steering test to determine the steering characteristic
constants K', Ti', T2' and T3'.
both steering tests mentioned above are also applicable to the phase plane analysis
and this analysis can be used to determine the reasonable characteristic constants together
with the frequency response analysis.
the steeriñg transfer futiction Y(iw), obtained from the frequency response analysis
in the intermediate frequency range provides useful data in connection with K'T3'/Ti'Tj' to
determine whether the course keeping ability, especially in the case of the automatic steer.
ing, is sufficient or not.

Introduction proper course-keeping quality of the ships.

Highly centralized automatic control sys- In fact, many super-large ships showing
tem of ships has been progressing rapidly, course-instability are operated sufficiéntly
accompaniéd by successive appearance of enough with auto-pilot systems, therefore one
specialized carrier and shortage of seamen. of the ship design problems is to find out the
Along with it, course-keeping quality be- allowable course-instability limit and con-
comes the center of interst in the field. of sequently the least rudder area neçessary
study on rnänoeuvrability of ships, especially for a given ship.
the relation between the course-keeping In general, the characteiistics of the re-
ability of ships with control systems in. sponse behaviour of a ship to steering can
corporàting automatic steering device and be well represented by four characteristic
constants K, T, T2 and T3. Usually, the
'

Kobe University of Mercantile Marine simplified steering quality indices K and T

149
150 Keiichi KARAStThO

are widely used. These indices together in practice. Further, the comparison between
with the results of the spiral test turned out the data of full-scale expêriments and, model
to be the common method of representing experiments is possible, and useful informa-
the manoeuvrability of ships. Hitherto, tion can be obtained in regard to the cor-
Numerous results of full-scale manoeuvre relation of full-scale ships and model ships
test to steering have, been summarized ac- on manoeuvrability.
cording to this simplified indices in the J apa-
nese shipyards, and these indices, K and T Principal particulars of given ships and
become the criteria to judge the manoeuvra- the 'test conditions
bility of ships. Recently K and T were used Full-scale experiments were carried out in
also for the calculation of ship motion caused cooperation with some shipyards, Japan Ship-
by automatic steering. However, when the building Research Association and using a
ship is steered frequently by manual or by traning ship of Kobe University of Mercantile
automatic steering and the course control is Marine. Results of the experiments on eight
taken into account, K and T are inconvenient ships, including the traning ship "Fukae-
for the analysis and synthesis of the ship's maru" of 360G.T., a car carrier of 2,600G.T.
motion, and all the four constants K, T1, T2 up to a tanker of 220,000 D.W.T. are shown
and.. T3 must be used. below. Four ships were fully loaded, and
This inconvenience is attributed to the the other four were in ballast condition.
fact that the motion mentioned above sur- Particulars of these ships and the test con-
passes the applicable range of the first order ditions are shown in Table 1.
system approximation in which the manoeu-
vrability of ships are represented with just Parallel shift manoeuvre experiment
a pair of indices K and T, and especially 2.1 Procedure and measurement in 'the
due to the negligence of the stabilizing effect parallel shift manoeuvre experiment
of the term K- T3á whiçh effects largely the In the parallel shift manoeuvre experi-
ship motion. Here, it is necessary to use ment, a ship is steered so that the eventual
complete representation of the response be- course after the steering is parallel to the
haviour of ships to steering with the four initial course. The resulting ship's track is
characteristic constants K, T1, T2 and T3 shown in Fig. 1. Considering the applica-
described above.' tion öf two types of analysis technique de-
The aim of this study is to determine the scribed later on manoeuvring experiments,
constants K, T1, T2 and T3 through full-scale combined use of manual and automatic steer-
measurements of response behaviour of ships ing was taken, in which at the begining of
to steering, other aim is to apply the new the experiment the ship was steered manual-
procedure of manoeuvring model experi- ly and then the helm was switched to the
ment'1 proposed by the author et al. to actual automatic one halfway. In carrying out
ships to examine its practicality. Besides, such an experiment, special attention should
the determination of the four characteristic be paid to start the experiment with cau-
constants was also attempted by the use of tious course-keeping on straight course by
phase plane analysis"'3 which application best performance of the automatic steering
has started in the treatment of naval archi- or by skillful helmsmen so as to insure non-
tecture problems.. Other advantages in this initial disturbance, also the course setting
experimental procedure are that there is no for the automatic steering equipment must
loss of ship passage and the chance to en- be done accurately o that the course after
counter with obstacles is small. It may be steering become exactly parallel to the one
said that this is the only transient response at the start of the experiment.
technique applicable to full-scale experiments An example of the procedure in the parallel
 Full-scale Measurement of Frequency Response Characteristics of Ships in Steering 151

Tabla i Particùiaro of the ShipG and Tact Condjtjonc

hipNuabor ..). 2 4 .6 '7--8 9

Kind of Ship Training Oil Bulk Bulk Car Ore Oil Ore/Oli.
Ship Tanker Carrier Carrier Carrier Cariier Tanker Carrier
Length Between Perpendiculars 37 0 300 0 162 6 162 6 115 0 2490 307 0 251 0
L(m)

ii,
Length/readth, LIB 4.75 6.0 -
6.55 6.55 7.-1 5.94 6.36 6,15
Breadth/Draught, D/d 2.89 2.64 2.72 2.72 3.24 2.95 2.49 2.69
Block Coefficient, Cb 0.564 0.030 0.790 0.790 0.526 0.823 0.850 0.821
L.C.B. from Midohip, 1.cb. O.95%a. 368%f. 3.96%r. 3.99sf. 2.90$a. 270%f. 2.72%f. - 3.l0%f.
Number of Eudder
Typo of Rudder
- - 1
Ordinary
1
Ordinary
- 1
t4arincr
1
Mariner
1
Mariner Ordinary Ordinary
- i
Ordinary
Rudder Area Ratio, A/L.d 1/34.15 1/61.8 1/55.1 l/55.11/44.3 1/61.7 1/69.1 1/66.7.
(with Horn) ith liorn)ith Horn)
- 1/66.4 1/66.4 1/59.2
(movable) (movable) Qaovable)
Rudder flight, h() 2.105 12.30 7.00 7.00 - -

lumber of Propeller 1 1 1 1 2 1 1 1
Propeller Diameter, O(a) 1.95 8.60 5.64. 5.64 3.285 6.90 8.40 6.50
Propeller Pitch Ratio Control 0 709 0 819 0 819 0 885 0 677 O 677 0 754
Number of Bladea. 3 5 5 5 4 5 5 .5
Direction of Rotation Left Right Right Right R,&L. Right Right aiht
Loading Öoniion Ful Full Ballast Ballast BaJ.1át Bàllhdt Pu11 - Ful1 -
Dioplacement (ton) 454 243 954 17 809 17 405 3 810 66 055 246 500 126 300
Tria, (d-df)/L 1.62%. 0% 2.35% 220% 1.21% 0.21% 0% 0% - -

¡lean Draught, d,,,(m) 2.70 l9O2. 5.57 547 - 3.999 7.854 19.35 14.63
Ship Speed, V0(kt) 11.5 17.5 16.4 16.7 19.0 15.0 15.8 16.5
Remarka . equippc CutUp - 5amc Venu Lirge &
with .torn . Lou9
Bow- :- -
.
Super
Thruotar S.No3 StrucLurc -

¡(t) rudder anIe

s heodin
turn on CutomQtiC steerin

- g
/
I i
,
//
I ¡

I..
t I

stèering for t
stroight course ¡

j manut steering
L autômotic steering

Fig. i Illustration of a Parallel Shift Manoeuvere Test

shift manoeuvre experin-ient is given belbw. for full-scale experiments.

1) Advance- straight ahead as exactly as Order 50 starboard with the helm at
possible by manual or automatic steering. the instant of the zero yaw-rate mentioned
Although in this state the ship advances in prócedure (1)
straight ahead, the yaw rate of the ship Order 50 pott with the helm when
varies around zero though slightly, but ex- heading deviatión reached at 30
periment must begIn with the ascertaining Order 5° starboard with the helm again,
condition in which the yaw rate is perfectly at 1° prior to approaching the original head-
zero.. Besides, it is convenient to take shIp's ing. Then, after overshooting the original
heading as round number of gyrocothpass as headslightly, the ship's heading starts to
152 , Keiichi KARASUNO

return to the original one again. on the parallel shift manoeuvre ex-
5) Turn on the automatic steering device perirnent'
to keep the original heading, just before The relation between the ruddér move-
reaching the original heading, usually Ö.5° ment 5(t) and the ship motion ç(t) in the
before. Course-keeping with an automatic linearized motion of the ship's manoeuvre
steering device is held up to approach to a can be represented reasonably by the equa-
regular damped or sustained yawing even- tion
tually. Moreover it is desirable tJiat the rud- d2ç1 dçt
der angle under the operation of the auto- TiT2--+(Ti+ T2)--+ '

matic steering device is always proportional

to the heading error, in addition continuous =Kô+KT3. (1)
active rudder movement is carried out .so as
to make the sustained yawing motion'. Expressing Eq. (1) using the transfer func-
For adjusting the automatic course-keeping tion for the rudder movement as input and
device, it is desirable that weather adjust- the ship motion as output, we have
ment, dead band rate control and so on äre b()
made as smalla s possible, and that the pro- K(l±PTs)
Y(p ) (l+pT1)(1+pT2) ç1(p)
ô() - 5(p)/p
portional steering of maximum helm adjust-
ment is carried out. Ptçíj(t)dt
'The above procedure is represented gra- - L t
- rI et \ 5(r)dr. dt (2)
phically in Fig. 1, but it is to be stated that LI \ 5(t)dt I

the experiments described in the present LJ0 ' J Jo Jo

paper must not necessarily follow the pro- where =a+iw represents a complex num-
cedure described above, instead, various ber.
procedures were attempted, such as the test In case of course-stable ships, both inte-
in which the automatic steering devicewas grals of the numerator and the denominator
turned on in place of procedure 4), test by converge with a-+O+, therefore we can write
bang-bang steering repeating procedure 4) K(1+iwT3)
and test by the automatic steering using it Ys(zw)-
(1+ zwTi)(1+ zw T2)
from the outset. The necessary time for any
of these experiments is usually about 15
minutes. =lim 5«t)e'te-1'tdt
O
(3)
,--(l+
The rudder movement 5(t) was measuréd 5 3(r)dr (It
with a potentiometer rotated by a piano 5

wire wound around a rudder stock, while Measuring /(t) or (t) for an arbitrary
yaw angle .çb(t) was measured with a direc- rudder motion 5(t) and substituting their
tional gyroscope, both were recorded by values in Eq. (3), the transfer function Ys(iw)
oscillographs. Measurement of yaw rate us- can be determined and then the response
ing rate gyroscope was also employed partly. characteristics of ships can be determined.
The yaw añgle and the yaw' rate Were The importance of using Eq. (3) in deter-
caiibrated.by reading heading angle of the mining Y in the range of small w from the
ship's gyrocompass. The ship forward veloci- parallel shift motion has been described in
ty was measured 'with a log, and the change literature', because no. eventual change of
in the forward speed was hardly observed ship's heading is caused.
except the parallel shift manoeuvre experi- N11Tic calculation of Y(iw) is per-
ment ,with, 1arge motion. formed for several values of w according to
Eq. 3). But it is difficult to carry out the
2.2 Outline of frequency response analysis perfect parallel' shift manoeuvring experi-
 Full-scale Measurement of Frequency Response ;Çharacteristics of Ships in Steering 153

ment in full-scale test, because of the ex- +(Ti+T2)çi0+T1T20

.istence of external disturbances acting upon
the ship in sea. It 'is therefore necessary AccordIngly, this value is subtracted from
measured '.
that some corrections are applied to the re-
sults of measurements1, such as corrections 2nd correctioncorrection of rudder neutral
för initial motion, rudder neutral position position error ¿5f'
error, residual steady turning, imperfect The apparent rudder angle -ô, measured
parallel shifting of the truck, etc. Thus the is converted to the true rudder angle ô in
result. may be converted to that for. the ideal Eq. (3) (ô=Òm+ô). The ships advance straight
parallel shifting, then the calculation accord- at zero true rudder angle.
ing to Eq. (3) can be carried out. Namely, 'D=5rt is 'subtracted from the
measured D
Improvement of correction'calculation
2.3 3rd correctioncorrection of residual steady
in the frequency response analysis turning ç
The outline of the frequency response in full-scale experiments 9L is expected to
analysis in the parallel shift manoeuvre ex-
.

become zero, because steering in this pro-

periments is as described before. In case cedure is carried out on the bàsis of the
of small amount of, the correction due to the ship's heading measured with gyrocompass on
imperfect parallel shift motion, the error in board: Sometimes a case. occurs, however,
the. correction using the indices K and. T in Which small is present dUe tò charac-
described in literature' may be small. In teristics of the. automatic steering device in
full-scale experiments, however, the time operation during short time of measurement..
constant T is large in general and as a re- At that time, stepped steering ô5-i(t) induc-
sult the amount of this correction is often ing the residual turning rate çii,,, and -its
not small .but. large amount of correction is motion response -.
required even forsmall initial motion. There-
D=t
fore the application of the first order system
approximation to this. case is improper, and
exact correction with K, T1, T2 and T3 is
(TT .et/rj
desirable. In making this correction, the (T2 Ta)
four characteristic constants assumed before-
hand can be òbtained either by finishing the
frequency response -analysis. or by the use +(Ti+ T Ta)ç&,ç&,.t
of phase plane analysis to be described later. are added to the measured D and .b values
The corrections based on K, T3, T2 and T3 respectively. -

turn out to be as follows, 4th correctioncorrection of the imperfèct

ist correctioncorrection of the initial mo parallel shifting of the truck çbm
tion-
-

Courses before and after the parallel shift

Assuming ço and ço as,, initial yaw rate manoeuvre must be made exactly in parallel,
and yaw acceleration respectively, which the but eventual change of, ship's heading .about
ship has at the, time of begining the test, 0.5° is inevitable in full-scale experiments,
"memory" of this initial motion is con-
which is caused by error in. course-setting
tained in ship motion recorded, it comes to and dead band and weather adjustment
be
of autopiiot mechanism. Besides, sometimes
=fi TIç/o+TlT2o
-
C-i/Ti these. parallel .courses may not materialize
due to the ist and 3rd corrections.
T1 T20 as time avarãge of
+T T2q0 +
T1T2
Now assuming çb,1
ship's heading measured from a base line
154 .; : :
Keiichi KARASUNO

after the ist-.. and 3rd corrections are In the case of full-scale experiment, some-
applied, correction is carried out by. adding times a case of ICI>1 may occur. When
impulse steering just sufficient for eliminat- C>i, then 1/K can be converged to a proper
ing this imperfect parallel shifting, and its value if subsequent approximate calculation
motion response assuming [Ke+ Y(0)]/2 as new Ke is carried
Dçbrn,/Ke out. When C<i, K comes to be zéro by
this method, and reasonable valuê of K can
/ (TiTs)çb not be, obtained. Accordingly it should be
T1T2 decided as satisfactory when there is no
(7'2 T3)çbmo, much difference between Ye(0) and presumed!
T1T2 K attained by trial and error method. Be-
Summarizing the above four corrections, sides, the situation of such ship motion of
DDrn,o(ör+«3s)t
C<-1 occurs when the initial yaw rate is
high and, in addition, hasthe same sign as
{D0+(T2-73)ç) yaw rate given by the first swing of the
parallel shift manoeuvre. It occurs also
T3 when course error before and after steering
{T2o+(T2--T3» is large, and the eve'ntual mean ship's head-
ing at the steady sustained yawing has añ
TT1T2o}]e/T1 opposite sign to the ship's heading given at
the first swing of the parallel shift motion.
T1o+(Tj We should pay specially attention to avoid
thses situations shown in Fig. 2.
T3 ¿ Tio+(Ti T3)
Example 1
bose line after
(I, No.123orrectiôns
T2o)] e/T2 + Ç& ± ç -t

where V

=no.±(ç1açi')(Ti+ T2 T3) Example 2

+ç1oTiT2+T3ç1'o, Dm,,=çbw,o/Ke 1'

are the base lines for çb and D. Substitut-

ing valus çb and D into Eq.. (3), Y5(0) can be base line
No.1.2.3 cor
computed.
5th correctioncorrection due to the dif-
ference between assumed K and computed Fig. 2 Examples of Erroneous Ship Motion at
Parallel Shift Manoeuvre (C<-1)
In general, final K.can be determined aç-
cording to the equation, When the exact correction based on the
above described characteristic constants K,
--=(l/Ki + C/Ke)/(1 ± C) T1, T2 and T3 is carried out, error due to
Where the corrections turns out to be less, and re-
asonable transfer function Y can be ob-
ICI <i, - t/ (2. dt) tained.

çb: yaw angle after ist., 3rd and 4th 2.4 Results of harmonic analysis on fulI
correctiòns V
scale experiments V

Ki: computed Ye(0) V V

V
Resùlts of the patallél shift manoeuvre ex
 Full-scale Measurement of Frequency sponse. Characteristics of Ships in Steering 155

periments and. its frequency response analysis harmonic analysis on the parallel shift mo-
was carried out for seven ships, namely S. tion is shown in the form of a Bode diagram
Nô. 1, 2, 3, 4, 6, 7 and 9; are shown below. in Figs. 10, 11, 12, 13, 14, 15 änd 16, and
The procedure of the experiments for each plotted with O marks in these figures. The
ship put to the test is different in detail.
be time histories of each rudder movement S No.4

ô(t) and its ship response çL(t) is shown in

Figs. 3, 4, 5, 6, 7, 8 and 9. The transfer
function Y'(iw') obtàined as the results of
5.- S.No.1

Fig. 6 'Record of Parallel Shift Mänoeuvre

sec.,
5 Test of S.N. 4

Fig. 3 Record of Parallel Shift Manoeuïie

Test of S. No. 1 Fig. 7 Record of Parallel Shift Manoeuvre
l'est of S.No. 6

-50rn

Fig. 4 Record of Parallel Shift Manoeuvre Fig. 8 Record of Parallel Shift Manoeuvre
Test of S.No. 2 Test of S. No: 7

.5.Ño9

/ Sec
/

Fig. 5 Record of ParàIIel Shift Manoeuvre. Fig. 9 Record of Parallel Sl'ift Manoeuvre
Test of S. No. 3 Test of S. No. 9
156 Kêiichi KARASUNO

02.
SN0 I
1og,lYl log,IYI
l- .0 - -
Frequency
o
Response
Frequency Analysis
Respoñse for Periodic
olysis Steeriiig
for Parallel
-li hifI Mono.yr--I
-
th tog.W !Q_ logw
-0.5
k1.47
T;3.10
T;1.06
K.O.8B 1.94
T.05l.0.52i
T0.51 -052i
T0.64
+ Frequency
ReSponse
Anolysis

I..
br PeTrioalc
Steering

.. 1.5 ioc! 1.5

Fig W Steering Transfer Function of .S. No. i Fig. 13 Steering Transfer Function of S. No 4
as obtained by Frequency Response as obtained. by Frequency Response
Experiment Experiment.
0.2

0.5 - log.lVl
Frequency
Respörtse
log .lYl
SNo.2
0.
Frequency
Analysis.
for rttel Response
SI-if t Manoeuvre
Analysis
8 for Parallel
-10.
o d'. log.W Shift Manoeuvre.

K2.09
T5.93
T0.76
Tl.7B
50--0.5
+ Frequency
Response. 8 +Frequency
Analysis Response
for Periodic Analysis
Steering for Periodic

-10-- 1.0
Fg. 11 Steering Transfer Function of S. No. 2
a obtained by Frequency Response
Steering

iM -1:5

Fig. 14 Steering Transfer Functiôn of S. No. 6

.:.
Expfiinent
as obtained by Frequency Response
Experiment
0.2 0.2
S No.3 SNo7
log.lYl log ,lYl
o o
Frequency Frequency
Response Response
Analysis. o Analysis
for Parallel for Parallel
Shift Monoeuvre Shift Manoeuvre

o 0--Q5 5
log.w

lVrl.1 6
T29
T;0.6.4
Iiif29Q
+ Frequency + Frequency
Respanse Response
Analysis Analysis
for Periodic for Periodic
Steering Steering

-1.5 -bc! -1.5

Fig. 12 Steering Transfer Function of S. No. 3 Fig. 15 Steering Transfer Function of S. Nô. 7
as obtained by Frequency Response as obtained by Frequency Response
Experiment Experiment
 Full-scale Measurement of Frequency Response Characteristics of Ships in Steering 157

o
0.2
S:No.9 tenon for judging up to what extent of fre-
log,1Y1
oTh 0
quency this analysis can be reliable. In this
Frequency
Response
figure, the ratio of the harmonic component
Analysis
for Parallel o
of D at frequency w to the zero-frequency
Shift Manoeuvre
q)
o harmonics of D is taken as thé ordinate, and
Iog,.Lv'
-10 -.Q5
cf- 0.5
05
W 136
o its change with the frequency . logiw' is
T;a89 shown.
T,,1 24
The outline of the results of each parallel
shift manoeuvre experiment and its harmonic
+ Freei
Response
analysis for each ship is described below.
Jialysis s. No. 1After the rudder was put to 3°
for Periodic
Steering port and then to 3° starboard by a helms-
man, the parallel shifting was carried . Out
Figé 16 Steering Transfer Function of S. No 9 by turningon the automatic steering device
as obtained by Frequency Response (Fig. 3). The results of experiments of the
Experiment same procedure repeated. four times are
shown in the form of Bode diagram in Fig.
Bode diagram reptesents common logarithms 10 including re-experiments held three days
of the noñdirnensional amplitude ratio later. The variation Of the amplitude ratio
IogioIY5'(ïw')j and the phase lag ç1Arg I Y3'I for each experiment was found to be
Ys'(iw') on its ordinate. Hereupon, w' mw/( VIL), within ±13% of their mean value, but the
Ys'(iw')m Y(iw)/(V/L), where V is forward phase lag was found to be in good agree.
speed. and,. L is ship length '
ment. The analyzed frequency extends over
Moreover, Fig. 17 shows ali example of the range logiow'O.l, and it was found
harnionic components comtained in tïre ne- that the harmonic component contained in
gral of the rudder motion Dm )(1)dt' in the ship motion is larger as the frequency is
parallel shift manOeuvre arid give some cri-
5
lower. The reliability of the results at low
frequency range increases, as canbededuced
10 from the diagram of the harmonic component
contained in the integral of rudder angle
ID(k')I/tD(o) (Fig. 7). Besides, in this ship,.the tendency
8- where ' of the, amplitude ratio lYs'! to decrease with
7
the decrease of the frequency. was observed
at the low frequency range, but it. seems to
6
be a special case in view of the previous
5 res ults.
S. No. 2After steering to ° port and 50
starboard respectively, automatic. steering
.

3 device was. put in action. (Fig. 4). Experi-

ments 'of the same procedure were repeated
2
five times, and the results are shown iti Fig.
11 In these experiments the initial motion
were different slightly from each other. The
o
-10 -o 5 O5 variation of amplitude ratio I Ys'I for each
Log,0il - (cvI..)
ex'perimentwas within ±20% of their mean
Fig. 17 Amplitude Components of. S(t)dt values, but' it can be said that the phase lag
against Frequency w' at Parallel Shift. was almost in. agreement. The resúlts cover
Manoeuvre the frequency range of logiow'O. The re-
:158 Keiichi KARASUNO

liability of the analyzed results are reflected are in good agreement with the vicinity of
by spectrum of rudder angle integral, and logiocu'=O. On the other hand, comparing
it can be seen that the reliability is best at the phase lag, we find that both q. values
logiuu'=O.l and decreases gradualy in lower are good in agreement at the low frequency
frequency range than that value. While in range, and there are differences of up
higher frequency range than that value, the to 130 in the viòinity of logiow'=O. The
variation of Ys'(iw') for each experiment is reasons of these differences seem to be that
very large, and the reliability of Y'(iw') de- the reliability of the analysis is low because
creases rapidly. of considerably large initial motion in this
S. No. 3After the rudder was put once experiment. Moreover, the accuracy of the
to loo port and thento 100 starboard, the experiment for S. No. 3 as mentioned before
parallel shift was carried out with course- is insufficient, and that there are some dif-
keeping by bang-bang steering of 50 rudder ferences in the magnitude of the motion for
angle. The. steering was carried out by a both ships. For all that the differences are
helmsman throughout, and the automatic too large to considered as éxperimental error
steering device was not employed (Fig 5). for two ships of the same type.
The analyzed results are shown in Fig. .12 S. No. 6After one cycle of triangular
for the range of logow'0.l. It was found wave-like steering of about 20° half ampli-
that the experimental points of Arg Ys'(iw') tude was given, the parallel shift was carried
at logiow'O.l are especially deviated from out by putting automatic steering in action
the smooth Bode diagram observed in thc (Fig. 7). The results of the harmonic analysis
common response characteristics, 'accord ing. extending over the range of Iogw'O are
ly the, response.' results at these points seem shown in Fig. 14. The amplitude ratio de-
unreasonable. This may be due to the effect creases with decreasing frequency in the
of inaccuracy of 'the analysis, because the low frequency range, similarly to the be-
course-keeping control in this experiment 'haviotir observed in S. No. 1. Besides, the
was made by the bang-bang steering, and steady yawing motion at the final stage of
also sufficiently steady sustained yawing mo- the parallel shift manoeuvre in this experi-
tion was notobtained. Besides, in the higher ment was short in time and was not steady
frequency range above Iogow'0.1, the har- enough, accordingl.y the period of the steady
monic component contained in motion de- yawing and transfer function of the fre-
creases rapidly, and it can be said that the quency are not exact. Moreover, the fre-
obtained Ys'(i') at this frequency range quency at which the speòtra of the rudder
lacks reliability. angle integral show maximum value is in
S. No.' 4After putting the rudder once agreement with the period of steady yaw-
to 100 starboard, the parállel shift manoeu- ing. As 'a result, it can be seen that judg-
vre was carried out with course-keeping by ing the accuracy of the transfer function in
bang-bang steering of 5° rudder angle (Fig. the vicinity of the frequency from the spec-
6). The analyzed results extending over the tral diagram, can lead to inaccurate deci-
range of logioco'O.l are shown in Fig. 13. sions.
In spite of the fact that the test conditions S. No. 7Very large motion occured, since
were almost same and the types were also rudder angle of 35° was employed (Fig. 8).
same for this ship and S. No. 3, the resu1ts of The test can be considered as an equivalent
the harmonic analysis on the ship motiòn to the so-called parállel shift measuring test.
was found to be in great difference. Namely, Results extending over the range of logow'
comparing the amplitude ratio I Y'j for both 0.3 are shown in Fig. 15. It was found
tests, we can find differences of up to 25% that the phase lag at logow'0.15 seems to
in the low frequency range, 'while both I Y/I be unreasonable, and the accuracy of Y'(ü')
 Full-scale Measurement of Frequency Response Characteristics of Ships in Steering 159

obtained in the frequency range higher than 3. Periodic steering experiment

that value is not good.
S. No. 9After steering once to 50 port 3.1 Harmonic analysis on trapezoidal steer-
ing experiment
and then to 5° starboard, the helm was
switched to automatic steering (Fig. 9), and The transfer function obtained Ys(iw)
course-keeping by bang-bang steering of through the parallel shift motion is especially
rudder angle was employed for the parallel lacking for reliability in the intermediate
shift motion (Fig. 31). Analyzed results of and the high frequency ranges. As experi-
repeated experiments for three times are mental technique to overcome this defect,
shown in Fig. 16 for frequency range of sinusoidal steering has been carried out in
logioco'0.1. The difference in Ys'(iw') for model tests. However, it is difficult in ac-
each experiment is large in the frequency tual ships to set the rudder in sinusoidal
range of loga/-0.6, and there are about motion, unless special devices are installed
±20% differences in the amplitude ratio j Y'j. on the existing steering systems. Accord-
In this respect, it may be one of the causes ingly, as an alternative solution, trapezoidal
that accuracy of analysis in low frequency steering is taken up, in which the rudder is
range on bang-bang parallel shift manoeu- changed port and starboard alternately with
vre is not satisfactory. On the other hand, definite rudder angle and period. When the
from the point of view of the accuracy of rudder movement ô(t) and the ship motion
analysis, it can be seen in view of spectra çi'(t) in steady state are expanded into Fourier
of rudder angle integral that the accuracy series taking exciting period as basic period
is maximum at logiw'0, and is de- and the first terms are adopted, the rudder
creases gradually in the lower frequency movement and the ship motion correspond-
range, while decreases rapidly in the higher ing to the sinusoidal steering can be obtained
frequency range, than that value of logiow' and then Ys(iw) can be computed. In ship
=0. manoeuvring, it is convenient to make the
Summarizing the results for the seven ac- average course of motion close to the origi-
tual course-stable ships described above, it nal course at the time of test, therefore
can be said that in general. the effective steering of the first half period is carried
analysis range of the frequency in the parallel out by being shortened to a quarter period
shift motion is generally logiow'0. Needless (Fig. 18). This corresponds to commence
to say, it depends upon the experimental the sinusoidal steering for model ships with
procedure. cosine phase. The amplitude of the rudder

¡(t): rudder ongte 'I'(t): Ships heeding

ç e. 5.
/ I.-
/

/ t..
/ I
I t

/ t
/
// /
t

s' / t

-5.
1,/2 1,/2 1,/2 1,/2
Tr14+
+ + + .
T;,2

Fig. 18 Schematic Illustration of Trapezoidal Steering with

a Constant Period
160 '..,Keiichi KARASUNO

angle is decided so that to keep the yaw frequency range. The-results of the har-
rate small enough to suit 'the linier monic analysis of the periodic steering shows
analysis, for example steering 'is carried out good agreement with the results of the
by taking the helm angle about '±3° in the transient response analysis on the parallel
low frequency range and ±10° in the high shift motion around log0 û/ 0.2, further,
frequency range. the amplitude ratio tends to decrease with
The above mentioned trapezoidal steering the decrease of the frequency in the low
is active periodic, steering in which the period frequency range.
is given beforehand, and the trasfer: function S. No., 2Results of the harmonic analysis
for any arbitraly frequency can be deter. of the modified zig-zag manoeuvre of 5°-1°
mined. 'On the other hand, the usual zig- and 5°-0.5° are shown in Fig. 11 at the fre-
zag manoeuvre and the steady yawing mo-. quency around logiow'0.1 (period of 170 sec)
tion in the parallel shift manoeuvre can be and link smoothly, with the r,esults of the.
considered as passive periodic steering mo- harmonic analysis in the low frequency range
tion from this viewpoint. When expansion ofthe parallel shift motion
into Fourier series is applied to these motions, S. No. 3Results of the frequency response
they can shoùlder a role in determination of analysis on the 50 zig-zag manoeuvre and
the transfoi- function Ys(iw). When the mo- the steady yawing motion in the parallel shift
tion is expressed by the measured yaw angle manoeuvre are shown in Fig. 12. The former
b(t), then ç is determined by differentiating is shown at frequency of logLo w' = 0.2, (period
the basic harmonics of the yaw angle 'after of 190 sec), and both amplitude ratio and
expanding it. intO Fourier series, and Ys(iw) phase lag give values close to Ys' for the
can be determined more accurately than the parallel sh,i'ft motion. The latter is plotted
method of directly determining 'ç. at the frequency of Iogioa'0.15 (period of
90 sec).
3.2 Results of harmonic analysis on full- S. No. 4Results of the harmonic analysis
scale experiments on the steady yawing motion in the parallel
A series of trapezoidal steering experi shift manoeuvre are shown in Fig. 13 and
ments by changing the period variously were are plotted at logo w' 0.25, thus these 'results'
carried out only for S. No. 1. As for other offer valuable data in the intermediate fre-
ships, the, usual zig-zag trials,, the modified quency range.
zig-zag ,trials4 and the steady yawing motión S. No. 6The results of the. frequency re-
of the parallel shift manoeuvre were con-' sponse 'analysis' on the 10° zig-zag manoeu-
sidered substitutes' Of the trapezoidal steer- vre are given at logow'-0.3 (Fig. 14). The
ing experiments. Results of. the harmonic motion is somewhat . small in comparison
analysis on periodic steering are denoted with the transient motion in the parallel
with + marks in Figs. 10, 11, 12,13, 14, 15 shift 'manoeuvre experiment, consequently
and 16. The results for each tested ship are the results are supposed to show poorer
described below. '

course stability than, that of the transient

S. No. 1The trapezoidal steering' experi- motion. On the contrary, the results come
ments were made. extending over the range to be somewhat more stable only by little
of O.4log1Od'O.4 (Fig. 10); this frequency margin.
range corresponds to periods from about loo S. No. 7The results of the harmonic
sec to 16 sec. In order to keep the' yaw analysis on the 10° zig-zag trial are showñ
rate small enough to suit the linear in Fig. 15'. The motion is quite smaller than
analysis, the amplitude of the rudder anglè that in the parallel shift manoeuvre experi-
was changed gradually from about ±3° in ment 'for the same ship. Accordingly the
the low frequency range to ±10° in high results show poorer'course stability than that
 Full-scale Measurement of Frequency Response Characteristics of Ships in Steering 161

of the harmonic analysis on the pararliel shift response characteristics with the four con-
mOtion in Fig. 15. stants K', T1', T2' and 7'3' using the transfer
S. No. 9The results of the harmonic function of Eq. (2). Ftom the above stand-
analysis on the 5°-1° modified zig-zag trial point, the four characteristic figures K', T1',
are shown in Fig. 16, and the results show T2' and T3' can be determined from the
good agreement with Y/ for the parallel shift characteristics of Ys'(iw')w'. obtained by the
motion at the frequency of logjo w' 0 (period experiments described above in low and inter-
of 190 sec). mediate frequency range by the use of Eq.
As a whole, the results of the harmonic (4) derived from Eq. (2).
analysis, Ys(iw), on the trape2oidal steering
trials, the zig-zag trials and the steady yaw- T1'+T2' i
-(w'7'3 cosbrsinç5r)
K' w'IYs'(iw')
ing motion in the parallel shift manoeuvre
seems to be in agreement with the results of T3'.T2' 1
the harmonic analysis on the parallel shift mo- K' - w'2-K' _W?hIYS/(iWF)lTa
tion having almost similar magnitude of mo- +cosçb)
tion to the periodic steering ones. Especially, (4)
it may be said that the trapezoidal steering,
Hereupon
which is the active periodic steering, make
up sufficiently for the defects of the parallel ,K'mK/(V/L) and T'1,2,3(V/L)T1,2,3
shift monoeuvre for the purpose of accurately The determination of the four unknown
determining Y3' in the intermediate frequency quantities K', T1', T2' and T3' is possible in
range which is important in the analysis or principle with four equations obtained by
synthesis of coursekeeping steering. selecting any two different value of w', but
actually more exact characteristic figures
4. Determination of the four characteristic are; obtained by applying the least mean
constants K', Ti', T2' and T3' sqüáfe method to Eq. (4).
The combination of the harmonic analysis In the examples of the test of the seven
on the parallel shift manoeuvre and the full-scale ships described in this paper,
periodic steering described in paragraphs 2 the frequency response experiments were
and 3 enable the determination of the trans- repeated only few times for every ship;
fer function Ys(iw) extending from the low Therefore numerical determination of the
frequency range to the intermediate fre- characteristic constant was carried out by
quency range. The characteristics of the selecting frequency range - as. L0-logw'
high frequency range is not important in the and by the use of experimental point. The
actual manoeuvring motion, and when the characteristic figures obtained are shown in
response characteristics in the low and inter- Table 2, and alsO the response characteris-
mediate frequency ranges are known, they tics Ys'(iw') computed with the characteris-
can be practicaly used in cases such as tic figures are drawn with full lines in the
course-control motiön by the automatic steer- Bode diagrams. These curves fit well with
ing and others. From the standpoint of the experimental points obtained from the
control engineering, it is not too much to harmonic analysis of the parallel shift mo-
say that only this Bode diagram is enough tion and the periodic steering motion.
to describe the response behaviour of a ship. Among these characteristic constants. ob-
On the other hand however, when the elu- tained above, time constants T' and T2' for
cidation of physical properties of the ship S. No. 1 and S. No. 6 were found to be con-
manoeuvring motion and its application to jugate complex numbers with positive real
naval architecture design are taken into con- part. This fact shows that the free motion
sideration, it is convenient to represent the of ships becämes damped-oscillating yaw
 162 Keiichi KARASUNO

Table 2 Steering Quality Indicies of Shies

Ship Frequency Response Analysis Zig Zag Test Tufning

No. r'm - K ' - T' 4-;-; T;!' +T.T3' I T' K'
1 0 035 0 88 - 0 64 1 04 0 84 0 38 0 85 0 59 0 92
2 0.037 2.09 5.93 0.76 1.78 0.82 2.54 4.91 - 2.00 3.45 11.0
3 0.097 1.16 3.29 0.64 2.00 1.11 1.05 .1.92
-

0.88 0.92 1.13

4 0.083 1.47 3.10 1.06 1.94 0.87 1.70 2.22
6 0.103 0.69 3.97 0.49 1.41 0.36 0.64 0.84
7 0 348 1 57 2 64 0 28 0 80 1 70 0 93 2 13
9 0.037 1.36 3.89 1.24 1.91 0.54 2.52 3.22

Ship Phase Plane Analysis - . - -

N O. Tm
/ K' T' i T' T' K'
Ti'T2'
T'12
T'-
Ta'
- -
+T2'-T3'
9 . 0.046 1.24 2.46 0.14 0.24 0.87 1.42 2.36
8 0.045 5.50 6.40 0.16 0.36 1.90 - 2.90 6.20

motion, but in'view of the numerical range. of the S. No. 1: it can not be said that the
of hydrodynamic derivatives constituting the peaks of j Y'j appeared for the same reason
characteristic constants, these phenomena as S. No. 1. On the other hand, in this ship,
can not occur in ships of normal types.5" Ta'. showed unusual large figures. in fact,
The results for each ship put to the test if the relation of (T1'+T2')2<Ta'2+27'l'T2' is
are described below, satisfied among T1', T2' and T3', the peak
S. No. 1-Starting from the phenomenon of Y'j appears at non zero frequency, and
that the transfer function Y' in the low fre- this ship satisfies the relation. in terms of
quency range increases the amplitude ratio the above two reasons, it is clear that the
with the increase in -the frequency, T1' ànd amplitude increasing effect of T3' is the cause
7'2' are calculated to be conjugate complex of the phenomenon that the transfer func-
numbers; -The above-mentioned sequence tion in the low frequency range increases
holds as far as T1' and T2' are conjugate the amplitude rätio with the - increase of
complex numbers: and in addition the con- frequency.
dition of (T'-fTi')<2T1'.T2' is satisfied no S. No. 7-Although the results here are of
matter how small T3' may be. the parallel shift with large mòtion, charac-
S. No. 2, 3, 4 and 9-T2' and T3' show con- teristic figures of T2' and T3' do not differ
siderably large figures in comparison with from the common practical ones.
common practical figures decided from the Summarizing the results for above seven
results of model experiments at fully loaded ships, it can b said that the combinhtion of
and even keel conditions. -
the frequency response analysis for the
S. No. 6-Though T1' and T2' areconjugate parallel shift manoeuvre and the periodic
complex numbers in this ship, but the cön- steering gives T2' and T3' considerablly large
dition óf (T'+'T2')2<2T1'T2' is not satisfied, values when the method of determinatin of
therefore these results differ from the results the characteristic - constants is according to
 Full-scale Measurement of Frequency Response Characteristics of Ships in Steering 163

Eq. (4) and the above mentioned experi- 3 and 9 are shown with o marks in Figs. 19,
mental results of Ys'(iw')w' characteristics. 20, 21 and 22, and the results of zig-zag tests
This fact requires through examination here- for S. No. 1, 2, 3, 4, 6, 7 and 9 are shown
after. Though the characteristic figures K', with o marks in Figs. 23, 24, 25, 26, 27 and
T1' and so on are important, from the control 28.
engineering point of view the determina- Comparing the values of K' and T'(m T1'
tion of the transfer function in the low and + T2' T3') derived from the harmonic
intermediate frequency ranges come to be analysis of the parallel shift motion and the
more important than those characteristic periodic steering, with those derived from
constants. Summarizing, it may be said that the ordinary analysis on the spiral or the
if only Bode diagram is available, the ma- zig-zag manoeuvre, it may be necessary to
noeuvring characteristics of ships are re- compare the indices K' and T' derived from
presented thoroughly. motions having equivalent yaw rate, since
K' and T' vary according to the magnitude
5. Comparative studies of the present results
with the results of spiral and zig-zag 5.No.3
0.3
trials
The results of spiral tests for S. No. 1, 2,
SNal
50
0.2

0.1
/
K_1.16
-15° -1 00
0 100 50
1

-0.1

-02

03
K': obtained from the Result of Frequency
- 0.1
Response Experiments
K': obtained from the Result of Frequency
Fig. 21 Turning Test Results of S. No. 3
Response Experiments
Fig. 19 Turning Test Results of S. No. 1 5.No.9
w'.
S.No2
r
0.3
o
0.2

0.1
j1 K2.09
00 -30° -20° 100
1100
20°5 300 4

7 nondimensionalized by Approach Speed 0h

) ': obtained from the Result of Frequency K': obtained from the .Result ofFrequency
Response Experiments Response Experiments
Fig. 20 Turning Test Results of S. No. 2 Fig. 22 Turning Test Results of S.No. 9
164 Keiichi KARASUNO

2
S.No.1 3 SNc.6

1 2
2

01 02 r.,. 03
Nondimensjonaljzed by Approach Speed
+: obtained from the Result of Frequency 0
Response Experiments (T'= Ti'+ T2' 0 0.05 01 r, 0.15
- T3') +: obtained from the Result of Frequency
Fig. 23 Zig-Zag Test Results of S. No. i Response Experiments (T'=T1'+T2'
T3')
Fig. 26 Zig-Zag Test Results of S. No. 6

S.No.7

o
O -I-

01 02 r., 03 o
+: obtained from the Result of Frequency 02 r, 04
Response Experiments (T'=T1'+T2' +: obtained from the Result of Frequency
Ti') Response Experiments (T'= Ti'+ T2'
Fig. 24 Zig-Zag Test Results of S. No. 2 7'3')
Fig. 27 Zig-Zag Test Results of S. No. 7

2 S,Na9
5.N3
o

iot' .,. o
K
o

O
01 02 r02
+: obtained from the Result of Frequency
Response Experiments (T'=T1'+T2'
o
Ti')
Fig. 28 Zig-Zag Test Results of S. No. 9
o
1 r, 02 of motion because of non-linear motion
+: obtained from the Result of Frequency of ships. In the present paper, the magnitude
Response Experiments (T' = Ti' + T2' of motion, the root mean square value of
7'3') non-dimensional yaw rate rm', is defined as
Fig. 25 Zig-Zag Test Results of S. No. 3 the average of extreme values of yaw rate
 Full-scale Measurement of Frequency Response Characteristics of Ships in Steering 165

appeared on the first two swings of transient K' due ' to the frequency respoñse
25). '
part of the paralleí shift motion ovér experiments is, however, slightly larger than
Accordingly, it means putting weight that due to the spiral and zig-zag tests,
on the initial part of the motion. The com- while '1" due to the former experiments
parison of the results of frequency respOnse turns out to be considerably larger than
experiments was made with K' of the spiral that due to the latter ones.
test results (mean gradient of r'ô curves) S. No. 6Though the cômparison' of the
and with K' and T' of the zig-zag test re- results from this approach with the results
suits obtained by extrapolating or interpolat- from the zig-zag tests is difficUlt, because of
ing 1/K', 1/T'rm' curves at the same r' to the difference in the magnitude of motion be-
the parallel shift motion. These values are tween the parallel shift motion and the zig-
shown in Table 2, and also the results ob- zag trial motiOn, it may be said that K' is
tained through the frequency response almost in agreement in both cäses. T1' and
analysis are recorded together with straight T2' according to the frequency response ex-
lines of gradient K' in the results of the perinients are complex number and show
spiral test (r'ö plane), and with + marks considerably small values in comparisOn with
in the results of zigzag test (1/K', 1/T'rm' ordinary ones by converting them to T'.
plane). ' According to the results of these S. No. 7The parallel shift motion is very
experiments: large and it is unreasonable to compare
S. No. 1The values of K' are almost in this results with the results of 100 zig-
agreement with the results of spiral tests zag test which has much w'eaker motion
(Fig. 19) and zig-zag tests (Fig. 23). On the than that, hut presiming from the latter
other hand, T1' and T2' are conjugate com- results daringly, K' derived from the parallel
plex numbers, but the comparison of the re- shift' manoeuvre seems to' be proper. On
sults of frequency response experiments with the other hand, T' due to the parallel shift
the zig-zag test results as for the tithe con- motion turns out to be rather large.
stants can be made by the adoption of T', S. No.' 9If the curves of the spiral test
since T'(T1'+T2'T3') comes to be a real results (Fig. 22) are regarded as intersect-
number. It is found, in Table 2 and Fig. 23 ing ordinate (r axis)' at five 'points, K' ac-
that T' derived from the frequency response cording to the frequency response experi-
experiments shows considerably smaller ments seems to be proper. However, com-
figures than those from the zig-zag tests. parison with this spiral tests is difficult be-
S No. 2K' due to the frequency response cause' of the insúfficiency of thé data of
experiments is almost in agreement with spiral tests in the vicinity of r'=O. Besides,
the results of the zig-zag tests (Fig. 24), above peculiar spiral characteristics ap-
while T' due to the former experiments is peared recently also' in a' certain model ex-
considerably larger than those of the latter periments7) and this was realized when a
experiments. Besides, K' according to the course-stabilized ship by so-called unusual
Spiral tests (Fig. 20) seems to be much larger phenomena turned out to be unstable due to,
than that of the ffequency response experi- the reduction of rudder area.
ments, but as the spiral characteristics at As described above, regarding K', the re-
sufficientlly small motion are not clear in sults of frequency response experiments and
Fig. 20, the comparison mentioned, above the results of other experiments seems to be
can not be made accurately.. almost in agreement generally, but the
S. No. 3 and 4It may be said that K' formers show rather slightly larger figures.
obtained in the present approach is in agree- On the other band, the values of T' calcu-
ment with the results of' the spiral tests lated through the frequency response. tech-
rather 'than that of the zig-zag tests (Fig. nique are considerably large. In cases when
166 Keiichi KARASUNO

T1' and T2' are conjugate complex. numbers, neglected, and Eq. (1) was employed. Now
T' dèrived from the frequency response ex selecting k points in time in a region in
periments are found to have small values which the rudder angle was kept constant,
than that from the zig-zag tests. As for and writing conditions of the mçtion as ,
this discrepancy, there are some problems Çik etc,
in the method of defining mean magnitude of T1T2 T1+T2
motion in the parallel shift manoeuvre, but ik+ cik+Çi,ikömi±ôr (5>
it should be remembered also that T' T1'
+ T2' Ti', which is the,promise of first order
system approximation, may. not be totally hereupon, 3,: measured rudder angle
correct. When importance is placed on values 5,.: correction of neutral rudder
of the tiansfêr function in the low frequency angle (unknown)
range, T' can be obtained as T1'+T2'T3', ôôm+ör
while when importance is placed on thehigh Besides, suffix .i indicates the region z in
frequency range, T' becomes T1'T2'/T3'. As which the rudder angle was kept constant.
shown in Table. 2, using the latter T', the Taking dfdçi. into consideration, the
agreement between the results of the pre- conditions of motion (çb, ç, ç,. öm) can be re-
sent approach and the results of zig-zag presented on phase plane of ç.çt by the use
tests seem to be improved slightly.. of the time t as a parameter (Fig. 29). Sub-
stituting the various values read on this
6. Attempt to determine characteristic con- phase plane loci into Eq. (5), T1T2/K, (Ti
stants through the parallel shift mono- + T2)/K and 1/K can be determined by the
euvre by phase plane analysis methOd of indeterminate coefficient.
In harmonic analysis of. transient response
motion as.shown here, when it is subjected
to external disturbance even if partly dur-
ing a. sequence of motion, the disturbances
may afféct the ship motion and spoil the re-
sult and give undesirable tendency Qf de-
clining reliability of the result. This matter
is especially rèmarkable in ships with poor
cöúrse stability. I.n order to make up for
this defect, determination of K', T1', T2',
and Ti' was attempted by the application
of phase plane analysis2'1 to the parallel
shift motion which enters into appikation
recently to studies.on manoeuvrability.. There
are some possibilities in this method that the
influence of partial external disturbance'be- Fig. 29 çb Phase Plane Trajectory with Con-
comes comparatively small. Moreover, since stant Rudder Angie
this analyzing method is suitable for steer-
ing conditions, of repeated definite rudder 1) Determination of T1 T2/K, (T1 + T2)/K
and 1/K'
angle., it is applicable to the trapezoidal steer-
First, in order to eliminate' the unknown
ing test also. neutral rudder angle, qs. (5) for different i
6.1 Analyzing method. values are put to subtraction,
First, it . was intended to consider such T1T2 ... ... T1+T2 '
(çbik/'j)
small motion that non-linear terms could he K
 Full-scale Measurement of Frequency Response Characteristics of Ships iñ Steering 167

(6) and after steering respectively, the deter-

±($ikç1J)Ôrni Ônij
mination of T3. is carried out according to
Eq. (7). As for '1T2/JC, (Ti±Ti)/K, 1/K and
And the second i is represented by j. Also Sr, the values obtained in above procedure
because only one k is enough in the j region, (1) are employed; Besides, when this method
representatián can be made by the suffix j of selecting time is adopted, effects of the
only. calculation errors for 1/K and Sr on T3 are
Taking n=3, the coefficients. T1T2/K, (Ti small becaue the values of 2ô1dt and (t1
± T2)/K and 1/K can be determined by a t2) are small.
5 LI

system of three linear simultaneous equa- By the above calculation in procedures (1)
tion in three unknowns from adequately
different conditions of the motion, but it is and (2), T1T2/K, (T1±T2)/K, i/K, T3 and Sr
better to use the method of least mean are determined, and all the characteristic
squares for values of n thore than 3. At constants are obtained. Among them, com-
this time, for the purpose of improving cal- parison will be attempted for 1/K with re-
culátion accuracy of i/K obtained from data sults determined by another method.
3) Examination of 1/K
in the vicinity of çÛ=O, the determination of
the coefficients is carried out in the form of T1 T2/K, (T1 ± T3)/K, 1/K, T3 and Sr are now
to be known amounts, but i/K, as an un
T3T2 (k_J)± T1T2 (k-.J)+ i known, will be determined again according
K (çiçj) K (kçiJ) K to equation (7).
(Ôm5mj) Selecting ti and t2 as times when Çi, ÇbL2

=0 since [bj becomes maximum, Eq. (7)

k=1,2,3,---,n turns to be
by multiplying Eq. (6) by the factor l/(ç1ik TiT1.. 1
j). Besides, the unknown Sr can be ob-
tained, if necessary, by. substituting the ob- =52ô74t±Sr(t2_tI)+T3ôm1
tained T1 T2/K, (T1 ± T2)/K and 1/K into Eq. (5). (8).
Then, if T3 is determined, all the manoeurv-
ing characteristic constants turn out to be Substituting tjie results obtained in pro-
determined. In principle, .Ta can be deter- cedures (1) and (2) into .T1T2/K, T3 and ô,-,
mined from the conditions of motion during 1/K can be determined. Thern accuracy of
steering according to procedure (1) in the this calculation may be confirmed by wheth-
present discussion, but as the exact reading er this i/K differs largely from the values
of the conditions of the motion during steer- determined in procedure (i) or not.
ing, (sb, , ç1, 5m, im), especially Sm and ¿m, Moreover, in case of dS/dç0 (Sd/dçf)
is difficult, it is intended to determine T3 determination of T1T2/K and (T1.±T2)/K äc-
actually by the f llowing method. cording to procedure (1) is difficult, but in
2) Determination of T3 this case S= lIT1 or i/T3 comes óut
Definite integration is carried out on Eq. and an approximate value of 1/K can be
(1) from t=ti to E=ta, and dividing both sides determined by the use of Eq. (9) which geii-
by K, we get the equation erally holds true when ôm=const.
T2T2 .. T1±T2 . i T1T2 T1+T3 Ôm+ôr
K K
[ /1]L2______[/,]t2
1
T ¿j
K K
[S]0+ K - i
=s;5t+or(t2_11» Tl[S,]2 (7) _=_s(T1T2 ±s T1T2) dS T1T2
J
Selecting ti and li as the time just before (9)
168 '
Keiichi KARASUNO

At the tinie of executing the above de- attempted for the expetiments of S. No. 8
scribed numerical calculation, it is proper to and 9. The records of time histories of each
make phase piane first. Yaw rate ç parallel shift motion are shown in Figs. 30
and yaw accelaration ç required for this and 31. On ç phase planes, only the
were determined by averaging çb(t) measured conditions of ship motion 'with constant rud-
with rate gyroscope and afterwards by der angle are shown in Figs. 32 and 33.
numerical treatment method3. Namely suc- Below are the records of the time histories
cessive - seven values of çi' read at proper of motion and the details of analysis for the
time intervals were taken up and approxi- two ships,
mated, with a cubic expression, and then the S No. 8After steering twice by definite
ç1 and at the median point were determined rudder angle, the parällel shift manoeuvre
according to this cubic expression. This was carried out by putting the automatic
operation was repeated by shifting the time steering device in action (Figs. 30 and 32'),
intervals by one interval each. but the. parallelism of courses before and
6.2 Practical examples of phase plane after the steering was destroyed largely, and
analysis
in addition, the test was finished without
attaining steady yawing motion. Accord-
Phase plane analysis described above, was ingly the harmonic analysis was impossible

f'A \"t8
(deg).'
f-...'
/ \
/
,\ ----
// \
\,\!
(degIs'/
i r,
¿00 _---..-_..__
,-00
(sec)

¿
/
'i»

¿
AH
Fig. 30 Record of Parallel Shift Manoeuvre Test of S. No. 8

/
2
S No9

I..

(sec)
500
,

12

-16
F-H b--1
Fig. 31 Record of Parallel Shift Manoeuvre Testof S. No. 9
 Full.scale Measurement öf Frequency Response Characteristics of Ships in Steering 169

I and ô, the combination of part ® and point

© and that of part © and point© in Figs.
31 and 33 were employed and mean values
of both results were adopted. T3 was deter-

I.. -4 mined with points © and . Moreover i/K

was also determined by the use öf points ®
and ® and points ® and ©, and Was checked
up-
As the results of the above described analy-
sis, K', T1', T2' and T3' thus obtained are re-
corded also in Table 2. For S. No. 8, a
Fig. 32 ççb Phase Plane Trajectbry with Con-
stant Rudder Angle at Parallel Shift
straight line with gradient K' obtained by
Manoeuvre of S. No. 8 this approach is drawn in the spiral test re-
suIts Fig. 34, and 1/K' and lIT' due to this
- Sio approach are recorded with + marks in the
results. of the zig-zag trials of Fig. 35.

N8 1.0

.102 eg/s)

8 12

K 5.50

-2
.30° 4 00
-3 2

--5
Fig. 33, çb Phase Plane Trajectory with Con.
stant Rudder Angle at Parallel Shift
Manoeuvre of S No. 9
K': obtained from the Phase Plane Ana ysis
and then phase plane analysis was executed. of Parallel Shift Manoeuvre Test
The figures of T1T2/K, (Ti+ Ta)/K, 1/K and Ò Fig. 34 Turning Test Results of S. Nô. 8
were determined by the use of part ® and
point ® in Figs 30 and 32. Next, T3 was S.Noß
determined With pöints (D) änd CE. Besides,
for checking up 1/K, this value Was also
determined from points cA) and ® and points
/
®and©. o
S Nô. 9Bang.bang parallel shift motion
of.5° rudder angle, for which the results of 01 02 r03
frequency response analysis had been ob- +: obtained from the Phase Plane Analysis
tained (Figs. 31 and 33), was analyzed. For of Parallel Shift Manoeuvre Test
the determinatiön of T11'2/K, (Ti+T2)/K, 1/K Fig. 35 Zig.Zag Test Results of S. No. 9
170 - Keiichi KARASUNO

The results of S. No. .8 obtained by the intermediate frequency range obtained by

above described analyzing technique seem the frequency response technique may give
to show almost close values in comparison useful data for deciding course-keeping capa-
with K' due to the spiral test results. There blity on coursekeeping steering, espécially
are no means to make comparisons for T1', by automatic steering device, in relation to.
T2' and T31, however, it may be said that K'T3'/T11. T2'.
and T3' ate slightly too small in com-
parison with values according to common Acknowledgement
model experimental results and theoretical At the time of the present study, guidance
calculation results. For S. No.. 9, T2' and
obtained here show consi,derably small was given throughout by Prof. Kensaku
values in comparison with the frequency re- Nomoto of Osaka University, and hereupon
sponse resúlts, and are not very plausible. deep sense of gratitude is extended to him.
However, major constants K' and T' seem Also heartfelt thanks are offered to those
to have almost reasonable values, though concerned in each shipyard and the Japan
they show slightly smaller values than that Shipbuilding Reseach Association who afford-
due to the frequency response analysis. ed convenience for fúll-scale measurment
tests and data. Thanks are offered respect-
Regarding discrepancy between both re-
suits of the frequency response analysis and fully to Captain Keinosuke Honda and each
the phase plane analysis especially the dis- crew member who cooperated at the time of
the experiments on the traning ship "Fukae-
crepancy for TI' and T3', examination- must maru" and Minoru Hirota who whole heartily
be made on analyzing and. experimental
techniques. But anyhow, this phase plane gave data concerning manoeuvrability of the
Fukae-maru.
analysis may b an auxiliary means for the
frequency response analysis in the present
discussion at the time of determinig the four References -

characteristic constants K', T1', T2' and T3' 1) NOM0TO and KARASUNO: A new Procedure of
through the parallel shift manoeuvre experi- Manoeuvring Model Experiment, Selected
ments Papers from .LS.N.A. Japan Vol. 7, (1971)
) M. BECH and L. W. SMITT: Analogue Simula-
Conclusion tion of Ship Manoeuvres Based on full scale
Combination of the parallel shift ma- trials or free sailing model tests, Hydro-og
Aerodynamisk Laboratorium, Nov. (1968)
noeuvre and the periodic steering may be Y. TANAKA and K. KOSE: Phase PÌane Analy-
suitable for full-scale manoeuvre experi- sis on Manoeuvring Motion of Ships, Spring
ments, and it may be promising as a means lecture meeting 1970, J.S.N.A. Kansai Japan
to determine four characteristic constants (in Japanese)
K', T11, T2' and 1V on course-keeping mo- M. FUJINO and S. MOTORA: On the Modified
tion. Zigzag Manoeuvre and its Application, J.S.N.A.
Both manoeuvre experiments described Japan Vol. 128, Dec. (1970)
above are suitable also for the phase plane K. NOMOTO: 60th Anniversary Series Vol. 11-
analysis technique, and this analyzing tech- Researches on Manoeuvrability of Ships in
nique may be promising for reasonable deter- Japan, Çhap. 2, J.S.N.A. Japan, (1966)
mination of the four characteristic constants S. MOTORA: Course Stability of Ships, J.S.N.A.
Japan, Vol. 77, Jul. (1955) (in Japanese)
in parallel with the frequency response
M. FUJINO: The Results 'of Reversed Spiral
analysis technique. Tests of a Ship, reported at Japan Towing
3).. The transfer function Ys'(It'n') in the Tank Committee, 2nd Committee, Apr. (1971),