ARCHIEF
A NFW TYPE SPEED METE1
.PPLICATION OF THE !CARNA?7 V
TEX PHCMC1
TO THE D3IGN OF A SPEED 1'ETFR
Hi88Jflit*U
ST-UBA, DR.ng.
rectr.Techn1ca1 Reee&rch Laboratory,
NIPPON KAIJI rLOAI
EE{:II 4i
n
HH
.ScheepLo.
Technische Hogescho e1L J j? - l.I I I t I.e) a aiaa talaiTABLE OF COìTLS
A BSTRA C T
Page
i.
INTRODUCTION 1II .
VORTEX STREET TÀRTING FRCL i BOL'IN FL
4III.
ÂLIC
L IFITIC
OFKARAN VORTEX TiET T
?ED 1kT
ô_I i
Iv.
CYLINDER PRODUCING LAi' VOkThÀ STReT
10Iv-1.
3ectional form and size
10IV-2.
Length )f the square cylinder
10IV-3.
.nd plates and side p1tes
111)
End plates
112)
side plates
13
3)
Length of end plates
15V. MLTHOD FOR PICKI1G UP TIlE VORTEX -Ii 18
VI.
SPD INDICATOR AND TL
iCOiU)IIG
VI-1.
Outline of the research work
2)
VI-2.
Jetails of the indicator and the
rii
CABLJ FR A TOVJ::G TYPE SFEiI
2ik 25VIII.
TÒST RESULTE OF TIE SiEED METER 26VIII-1.
Distance of tbe preliminary
r'in ìt
a mile post speed
tri1
26VIII-2.
Error in the rnile-r'ost disterce
and the shallow water
effects
29viIi-3.
L:easìreient of sea waves
29VIII-3-i.
*aves ruade by the ship
30
wave length
30
wave height
31VIII-3-2.
Sea waves
32
VIII-3-3.
Informations
34hJIII4.
Lerith of the twing
cable
36I:L.
Ci:LLUI3N
40
It is of fundamental importance to measure accurate speed of a
ship when we investigate
more precisely into
hersea-going
per-forrnences. It Is also expected to become quite necessary to obtain
a reliable information
on
the ship speed from view point of a moreprogressed research on the strength
and
stability of ships, towhich the bydrodynamic action due to the ship's motion is
closely
related.
Be that as it may, most of the modern ships are
equipped with
speed meters of various types. These speed meters are calibrated
generally on the occasion of speed trials in a mile-post course.
However, it is very difficult to evaluate the
true
speed of a shipfrom such trial results which inevitably include influences of tide,
wind and sea wave. Moreover, even if the speed meter is calibrated
at a particular condition, for example, at a trial condition, these
calibrated resu1t are effective only for the specific condition
and not useful for any other ones.
This is because due to a fact
that the value indicated by an ordinary type speed meter is closely
related to the variation in trim and loading condition of the ship.
So far as the accuracy is concerned, lt has therefore been unable to
put full confidence upon a speed meter of usual type.
In order to solve these problems, the following two points
should
be considered; firstly, ;o obtain a speed meter
sufficiently accu-rate in itself, and
secondly, to find out a new method of getting a
true ship speed under any conditions required instead of performing
the mile-pcist course speed trials.
If a speed meter of high accuracy be available the latter
becomes a problem of minor importance, because a calibration of a speed meter attached to the ship can be made satisfactorily by means of a towing type speed meter of high accuracy working at a distance far behind the ship, where the local disturbances due to the ship hull become extinct and the speed meter indicates a true
value of the ship speed, ( the wave caused by the ship does exist,
however, its effect on the speed measurement can be excluded by
methods as described later ) . In this paper, therefore, the problem
on a speed meter of high accuracy will primarily be discussed.
Generally speaking, necessary conditions for an ideal speed meter can be summarized in the followings;
it must have high accuracy,
it must be free from such effects as wear, corruption and temperature variation,
constant relationship between the indicated values and the speed must be maintained, a linear relation through the origin being preferable.
2
Any one of the usual speed meters, such as the resistance type, whirling blade type, Pitot tube and heat wire types, does not
fulfill these conditions. The
author has
been trying to get anaccurate speed meter for many years, and finally noticed to utilize
the relationship between flow speed and frequency of
vortices
startedbehind a body running through a fluid. An extensive investigation
onthe trial production of this new type speed meter has then been
carried out,
and the results were reported in the congress held bythe Ship Experiment Tar
in Tokyo at the end of World War II.
Later, in 1951, some of the results appeared in the publication,
Tank Report No.?.
The first time 'when the new speed meter was
tested on an actual ship was on the occasion of the maiden voyage
of the single screw Cargo ship "NISSEI MAiTJ" from Yokohama to
Vancouver, where an investigation Into the sea-going performances of
ship was carried out from Dicember, 1951 to May, 1952.
Since then,
a large number of exDerlmental tests h've been conducted on speed
trials at sea in order to assure of a general use, succesive
im-provernents being made for obtaining sufficient durability and
satis-factory functioi of the pick-up as well as the recording part of the
speed meter.
Since lt is believed that the newly designed speed metei
can be used without any anxiety in an ship the results obtained from
the authors study will be presented in this report.
1
II VORTEX ST1EZT STARTING FROk
A BODY IN FLOW
Since the original Investigations on the vortices associated with
a moving body In flow was performed quite a long time ago, lt seems
very difficult to remember an exact date or person of the discovery
on the periodic phenomena of the vortices. The first experimental
observations may be due to STROUHAL who showed that the frequency depends on the relative velocity of fluid, and thereby the following
non-dimensional number being called STROTJHAL Number,
tj
V
where
n ::: number of vortices starting from one side of a body
in flow within an unit time
B :::: breadth of the body, V :::: speed of flow
RAYLEIGH also performed similar experiments, and he pointed out
that the STROUHAL Number S is a function of REYNOLDS Number.
In 1911, KABMAN presented his famous theory on the vortex street,
and the vortex street is then called KRMiN Vortex. It may be said
that no substantial advance in the theoretical treatment f the problem
, , ,
has been made since the KARi1ANs paper was published, however, a large
number of experimental works have been carried out for a further
de-velopment of the study. Experimental investigations by FAGE,
hOVASZNAY and H.BL1NKE, for instance, would be the typical ones.
4
From the results of these experimental studies lt can be
con-vinced that
there exslst
three cheracteristic ranges In REYNOLDSNumber,
namely,"stable"
C R.No. about 40 to 150 ),"transition" ( R.No. about
150 to
OO ), and
"irregular" range (R.No. about 300 to 10,000 ).
In the stable range,
vortices which
are caused alternately fromboth sides of a body are stable and quite
periodical to a point
somewhat far behind a body,
and
theyfinally
disappear due tovis-cosity of the fluid. In the
transition
range,vortices
areunstable
and sometimes it is even impossible to
find out
their periodicity.In
the
irregular range, vortices are not so stable as in the case ofthe stable range, however, a remarkable periodicity can again be ob-served, and they quickly disappear by the effect of the turbulence
in flow.
The STROUHAL Number,
increases
considerably with thein-crease In the
REYNOLDS NO. In the
stable range, while In theirregu-lar range it becomes almost constant irrespective of the REYNOLDS No.
In the transition range, the STROtJHAL No. is scattered and sometimes
even difficult to find the periodicity.
III APPLICATION OF Th CHARÂCThRISTJCS OF
KRMAN VORTEX STREET TO . SPEED METER
As can be seen from
the exp1antion In the previous article, the
STROTJHAL No. varies with the REYNOLDS No.
lfl
e very complicatedmanner.
In general, the
TFOUHAL No. S may
be expressed in the following;çconfiguration, R.No., roughness, turbulence
and quantity of gas desolved in the water
)If we neglect the effects of
roughness of the surface of a bodyas well as turbulence and
content of gas in flow when thedesolved gas
becomes separated in the flow behind a body, it givesn effect equivalent to e change in the form and
the size of abody
the sbove reiation becomes as follows;
( R.N))
In the case of a circular
cylinder, TYLER gave a formula,
=
.çq8(i
)for R.No. lower than 500, and concluded
that S becomesconstant when
R.No. Is higher
than 1O (
the maximum R.No. inhis experiments was
4x
fO).
According to the BLENKF-FUCHS-LIEEERs
test results, it is alsosaid that S increases until R.No. reaches about lO, and lt remains
constant, 0.202, in a range of
R.No.
between 1Oandwhereas 1f
1- t
R.No. exceeds 10 it shows a tendency of
decreasing sl1ghly but again
starts to Increase a little.
-6
j
Most of the experiments
conducted up to this
time bave been
onthe specimens of
circular cylinder, for
which it is considered that
as R.No. increases the position
of the tr1tion
point of the flow
separated from the body moves forward, causing
unstable fluctuation
in its position, and
when further increasing
the R.No. the
trans-ition point reaches
the laminar separation point on the body
(
i.e.
at the critical
REYNOLDS Number ), end
a so-called turbulent
sepa-ration starts to take
place.
since the viscosity
of the fluid
influ-ences del1ctely upon the characteristics of flow around the circular
cylinder where pressure
gradient varies gradually from positive to
negative along its surface,
the STAOURL No. should
very quite
complicatedly with a change
in the REYNOLDS No.
It seems, however,
that the characteristics
of Karman Vortex
Street
in a fluid would most
suitably be applicable to
the principle of
speed meter, provided
that one could find out a body of special
form
for which the
STROUBAL NO. becomesa constant value
Irrespective of
REYNOLDS Numbers.Before going into
a detailed discussion
on the
application of the
Karuin Vortex Street
to a speed meter, the
following general
considerations will
first be made.
The drag coefficient
of a body in flow,
behind whicI
rows of
vortices are caused, can theoretically be
expressed by e formula
cx=
2[ oJq36
0.3141where
1space between
two adjoining
vortices,
Bbreadth of a body
measured perpendicularly
to the direction
of flow,
u=advance speed
of vortices,
u0=-speed of a body.
8
As can be seen from Table i the c1cu1ated
values of C, which
are
obtained by using
experimental data of 1/B
and u/ut
(theoretical
ones being impossible to obtain
) ,coincide
quitesatisfactorily
, ,
with the test results of
C.
Then it may be said that, when KARiANVortex Street follows
the bodyin flow, the
drag coefficientC..
takes a certain particular value
depending upon
the characteristics
of the vortices.
In other words, the
experimental fact ofe
constant
drag coefficient
C. in a certain
range
of R.No. means that
, ,
the characteristics of KARMAN Vortex
street also have remained
unchanged in that range. From this
point of
view, itwould be
useful
to observe
the experimental results shown in Figs i and 2.
Figure 1
showing experimental date on e circular cylinderobteined
by E.F.RELF illustrates the relation of C.
and ' to R.No. As can
be seen from the
figure,the curve of
thedrag coefficient C
,
as
'well as of
v'
indicates a complicated
form with a variationof R.No,
and they show an abrupt change after
exceeding
a critical R.No,where the periodic phenomena of vortices can
hardly be observed.
In
Fig.2 shown are the test results on
a sphere and a circular disc as
illustrated in
Gottingen
Report, representingthe relation between
drag coefficient
and R.No.
It can be seen
from Fig. 2 that the drag
coefficient curve for a sphere looks similar
to that
for a circulercylinder shown in Fig. 1, while in the case of
a circular disc the
drag coefficient
is almost constant throughout the tested range Of
R.No.
( from 3</Q-.-i6
).
Considering these test results, the
author presumed that
there would
be no effect of viscosity on the separation of flow around
a
separation point of flow locates always at both eges
of the cylinder at any
values of R.No.
If
lt is so, the dragcoef-ficient or the S value which represents characteristics of vortices
would become constant when R.No.
exceeds a certain value.In order
toconfirm the
truthof
thepresumption, a
literature
survey on the experimental work was made. Nevertheless, none of the
test results on a triangular or square formed
cylinder was found,
since most of the experiments were conducted on circular
cylinders
although a few
canses on
flat plates or aerofoil shapes vere alsoreported. Only F.DURALL
explained in his
"Aerodynamic
Theory" thatin the
case
of a body havingsharp edges
nB/V wasconstant throughout
the whole range of R.No. in his experiments except at its low values.
It is therefore impossible to determine exactly
an upper
limit ofR.No. to which tile
nB/V value stays
constant, but it may be estimatedfrom the
above-mentioned testresults on a circular disc that
Sremains
constant up to lOof R.No.
Under these circumstances, tests 'were carried out on
triangular
cylinder, and
their results
seemed quitesatisfactory for applying the
characteristics of Karman Vortex
phenomena to theprinciple of a newly
designed speed meter.
( See the
Report No.? of the Ship
ExperimentTank in
Tokyo
) .Further
investigations were then made
on such
problems as determination of
a suitable form and size, pick-up system,integrating method,
recording
set-up, towing cable, etc.that must be
considered toward
its realization into a practical use.U
Iv CYLINDER PRODUCING KA1AN
VORTEX STREET
Iv-1. Sectional form and size
Although lt is not necessarily limited
to a
triangular or square
cylinder that can be
utilized for a
speed meter (see the previous
article ), a square cylinder
is currently used simply
because of itseasy production and
convenient fitting.When a required
speedrange for
measurement is given,say from
0.5 to 50 knots ( 0.26 to 26 rn/s ), the minimum size of the
square
cylinder should be deteruilned from the
lower limit of the
speed. If
we take
the allowable minimum R.No. conservatively tobe about
then the minimum breadth of a square cylinder is determined by the
following
relation; 1/2_)'
where B is
thebreadth of a square cylinder
measured
perpendicularly
to the direction of
flow (length of the
diagonal line) and Is
kinematic
coefficient ofviscosity.
Taking that V=O.26 rn/s
and
) =j.Aò( salt water at 15Th
),
becomes 2.23 ein, namely,
a square
cylinder of 1.6
cm in each side.
Thus, taking into
consideration
theconvenience for its fitting a square
cylinder of about 2
cm in each
side is now
used as a standardtype.
Thecorresponding range of R.No.
is from 6.3 - 63 x
IV-2. Length
of the square cylinder
It is said that
the end effect of
a cylinder upon the
flow is
re-markable within a
range of elength five times the
breadth, but lt
lo
diminishes at a distance of about eleven times breadth from the end.
Nevertheless from the test results on a square cylinder of a 1.85 cm
In each side ( see Photo. i
) , a regular periodicity of the vortices
could hardly be observed even with a voriation in the length up to
fifteen times the breadth, unless fitting end plates on both ends of
the cylinder. 3ince these experiments
were
carried out in aiming atan investigation of a log attached to the bottom of e ship without fitting the end plates, further tests were suspended in view of the
above-mentioned results. It is thus decided to fit end plates on
both top and bottom of a cylinder for any kind of speed meter Irre-spective of whether a towing type or a ship bottom attachment type.
Suiteble length of the cylinder with the end plates is considered to
be approximately equal to
twice Its breadth, but in the actual casethe length could be varied as long as a regular periodicity be
available.
IV-3 End plates and side plates
1) End plates
The Irregular motion of vortex rows that has been observed in the case of a cylinder alone would be
due not only to the end effect of
FLOA'
a cylinder but also to the
f
,
\
unstb1eness of vortex line as it
PLAT-R'ECTPAN(JULA CYLiNDER
4f
is not always straight along its S1tE PLATE
ength. To avoid these effects,
it is therefore
preferable to make the cylinder length as shorter as
possible and to fit end plates at both ends, thereby realizing an
(2)
12
ideal two-dimensional flow. The 8uitable distance between the two
end plates ( which Is equal to the length or a cylinder ) has also
been found to be about twice the breadth of a cylinder.
The effect of the
breadth of end plates upon the
characteristicsor vortices is not so
simple.
When we assume a source in place of acylinder which
Is located between two Infinite end plates, theve-locity potential is given by the following
formula,
a -' 2
)=
VZ
o9(x-)
sou i'CC
Then the velocity u at a position
of y=O Is given by
( e:
u=v1-., 3) ) from which lt can be noticed that
the effect of a cylinder upon the
flow diminishes inversely proprtlonal to the distance from the
cylinder.
If the breadth of end plates is finite, B, , it may change the
characteristics of flow
due to a disturbance dependingupon the
value of'tat both sides of the end plates. Assuming that this dis
turbance
be equivalently
replaced by another source at the sideof
the end plates, the effect on the flow can be expressed by a'/x'
Furthermore, with respect to the centre lt becomes
''/
In the case of end plates of B1
, it is given by 8',( L
f,
indicatingthat the effect of end plates diminishes very rapidly
with an in-.
crease in their breadth.
Consequently
if the end plates of aproper-.
ly large breadth be used, then their
effect will become of a negli-.
gibly small one. In other words, k value in the
following
ex-pression couldapproximately be taken as unity even in the case of
finite end plates a long as their breadth is
sufficient
large.s= V,x%Çr
It has been found from the author experiments that if the breadth
of end plates Is larger than four times the cylinder breadth, B,
then
the
rows of vortices become quite regular, n/V value beingconstant. However, since further relevant tests for obtaining an
exact value of Ç itself have never been carried out yet , values
in each case are now determined through a calibration for the
measurement of both n and V values by using a cylinder with end
plates.
2) SIde p1tes
Test results on the effect of the side plates as well as of enu
plates upon the frequency of vortices are
illustrated In Fig.3.
Itcan be seen from this figure
that the
frequency increases by about4% as a consequence of fitting side plates in addition to the end plates, of which the breadth is approximately eight times breadth of
the cylinder (compare the results marked i and 2 in Fig.3). On the
other hand, there seems no significant effect of a variation in e
distence between the side plates (see the test results marked 2,3
and 4 in Fig.3) It is thus obvious that the side plates influence
delicately on the frequency of the vortices, and therefore some
theoretical considerations will be made in the following.
Circulation of a vortex starting from a cylinder is given by
r = C
IT
dt
T=
,
C =
COiStQlt
(4)
Moreover, the following two conditions for KiBMkN Vortex Street
must be fulfilled;
-fl2V2 u.L
=COflStQflt
-Using the definitions,
-< L
v=vt
in case of end plates only,
v -= v0
in case of both end
and sideplates,
Equation (4) becomes as follows,
=
cÇïû1J0dt
a=*->1
7)
In the case of (B), vortices have a
tendency
of being attractedtoward a wall (side plate), that
means an
increase in hand
thereby
in i according to Eq. (6). This leads to an increase In
fl also
(see Iq. (5)).
When the distance between
side plates decreases'becomes large, however, it can easily be expected
that
there
exists
a certain range of the distance for which.T does not
vary so
signifi-cantly
due to the increase
inf . This would be the
reason why the
test results did not indicate
any
pronounced differeces in thefrequency in spite of the variation of the
distance between two side
plates. (
See
the test results marked 2,and 4
in Fig. 3)
Although [' decreases when side plates are removed
a=Yt/y.
becomesunity, and this Is so influential compared with
the decrease
in
circu-lation that T
must have Increased
(
i.e. n
decreased) in theabove-mentioned tests.
(See the test results
marked i and 2 in
Fig.
3 ).
From these
results it can
besummarized
that a linear relation-14
ship between n and V is always maintained, though in the case of e
cylinder with both side and end plates the ratio
t
differs alittle from th8t for a cylinder with end plates only. But the
ratio was constant irrespective of a variation in the distaìce
be-tween two side plates within thé tested range of Bg/B from 4.7 to
7.9.
3) Length of end plates
When the circulation of a vortex initiated along the edges of a
cylinder reaches a certain definite value after its gradual growth, the vortex starts to be separated from the cylinder and is shedding away in the after-flow.
It is defined
T=period of vortices generated in succession,
V=mean speed tangential to the surface of a cylinder, rdensity of circulations distributed on the surface
of a cylinder.
Since ¿r- is determined by the form of the cylinder and its advance
speed as well as by the distributed vortices In the after-flow, lt
can be
expressed by
C 8)
)1=o
where
¡;
depends only upon theform and the
advance speed of thecylinder, ì;,r3.- upon the corresponding
vortex
in theafter-flow. Then the
circulation
fl
can be given by the followingequation.
(A
-L0W
---*----In a case as shown ---*----In sketch (a),
forexample,
the third vortex andthe following ones locate outside of
end plates for which3X'--become smaller than
and
2 ' and thereby it is clear from-, ¿J _s4 l I J I t ¡,P I1 j 1111Ff i
-'
16
C
Çrd
= C(9)
2,7 = Cn
Li d t =
c:
Since the effect of a
vortex in theafter-flow upon
the cylinderis less pronounced with
an increase in the distance fromthe
cylinder,
clibecomes
smaller for a vortex far removed from thecylinder.
Considering a cylinder
located
between end plates, vortices inthe after-flow remaining
between
the pletes do not represent anyremarkable
changes, while
every one of those come out of theplates
form a vortex string which is
cut
off at both ends and a tendencyor growing into an infinite vortex. As the total circulation is
constant, the circulation in an unit
length becomes
ì1/,
, abeing the distance
between the end
plates, and 1,the length of a
Eq.(9) that T ahould increase (
namely, the frequency n should
decrease
)so as to keep
r'
constant.Figure 3 illustrates
the results of tests conducted for thepurpose of
determining a critical length of plates for which theef-fect of
vortices outside of the plates
upon
the frequency n can be1gnored
( See thetest results
marked 2,5,6,7 and 8 in thefigure.)
From Fig.3 the results are reduced to a curve in Fìg.4 where it is
seen that the critical length
is approximately 6.5 times the
breadthof the
cylinder.It is
also noticed that this length is almost equalto a space between adjoining two vortices in line calculated from
the test results, and thus the length of end plates necessary
for
Ignoring the influence of the redundant vortices upon the frequency
would be such that at least the first two vortices started from one
side of a cylinder are included within an inside of the end
plate.
( See sketch (b).)
From the results of the investigations as described in the fore-.
going it i decided
that
only endplates will
be attached to a squareu1
cylinder foc a towing type speed meter, while both end,slde plates
be fitted to
that of a ship-equipped speed meter.It is simply
be-cause that a ship-equipped speed meter must be used through a
cali-bration ( speed vs. frequency ) in order to
eliminate
effects of hull,and therefore smaller
one
becomes preferable as long es a linearre-lationship between
speed
and frequency is maintained.General
discussion on
the fundamental problemsconcerning a
speed
meter of this type has been made, and hereafter some considerations
into a detail of the
practical
problems such as type of pick-up,indicating and recording set-up and towing cable, will be
described.
V METhOD FOR PICKING UP
THE VORTEX FREQUENCY
At first sight it
seems quite simple to
pick up thefrequency of
vortices, however, there involved several conditions which have to
be considered
in the actual production of the apparatus.These would be;
high sensibility without causing
a mi8action,constant out-put for one cycle
irrespective of
the frequency,
simple construction,
durability, and
to be independent of
the water pressure
(
a pick-up of towing type sinks down when
ship stops ).
A method of picking up the pressure
differencebetween the right
and the left hand side on a back surface of cylinder'was first
re-garded as an ideal one, and
my efforts were directed to an
investi-gation on the
application of thesurface phenomenon of
mercury to the
system. But the
results seemed unsatisfacto'ybecause
its rnisaction
sometimes taken place in case of a high frequency could
hardly be
e-liininated and the pressure
differencebetween
each side ofthe
surface was not sufficiently large. After all it was concluded
that
the
most simple and suitable way is to pick up an electromotive
force
produced by e vibration of a movable flag
attachedto the cylinder
behind.
The first pick-up of this
type tested was of anelectromag-netic one and its mechanism
is illustrated in Fig. 5 (a).
It
as
found, however, that the
pick-up of this type was lacking in the
-19
bility
due to a wear or abreak-down of the flag
shaft, i.e. thevi-brator, which was sometimes experienc
as a
consequence of theex-istence of stoppers necessary for its mechanism.
Moreover
theout-put or the pick-up of this type is too sensibly inf1uened by a
slight variation of the vortex frequency, and its restricted
movement of the vibretor
due to the
stoppers might changethe
origi-, ,
nal characteristics
of KABIVIAN Vortex Street in theafter-flow.
Further
improved pick-up
system of anelectromagnetic type was
devised as illustrated in Fig. 5 (b). This new type of the pick-up
has the following advantages; firstly, since no stopper is attached,
the flag shaft is free from its break-down,
secondly the flag canfreely rotate to a wide range of angle so
that
it does not influenceupon the flow, end thirdly a deviation in the out-put corresponding
to one
cycle of the vortex frequency is
notsignificant.
As the
pick-up of this
type is of a simple
construction andfurthermore lt
has been found from the results of
actual applications that
itsfunction is satisfactorily reliable, the present use of the pick-up
is limited to this
type only.It should be pointed
out that there must exist
an optimum lengthof the
flag which responds mostcorrectly to the vortex
frequency.C
See Ref. 1 ).
From theresults of the
test on flags having alength of 0.35 to 2.21 times breadth of a cylinder, it
can be said
that a flag of 0.7 to 1.1 times
the cylinder breadth gives
satis-factory
results,and hence generally used is
a flag of which thelength is equal to the breadth of a cylinder.
VI SPEED INDICATOR AND ThE RiSC ORDING APPARATUS'
VI-1 Outline of the research work
The frequency of KABMAN Vortex Street as directly governed by the
advance speed of flow can be detected by the above-explained
e1ectro-magnetic pick-up which is connected to a cabtire cable leading to the
inside of ship. Then, by means of a
suitable appratus the frequency
of the vibrator may be converted into an exacty proportional
amount of electricity which exerts a speed indicator of a direct-reading
type or its recording system.
The apparatus should
satisfy thefollowing
requirements:The final out-put
corresponding
to an unit cycle must alwaysbe constant irrespective of either the magnitude of electromotive
force picked up or the wave pattern of each cycle.
It should not be influenced
by a possible variation in voltage
of power supply.
The mechanism should be as compact and simple as possible.
In general,
the magnitude of the electromotive force picked
up by
3 flag fluctuates
with the
variation of the advance speed of e
cylinder, and besides,
the wave pattern of the
electromotive
forcecorresponding to each
cycle of the vibration
Is not necessarily
the
sanie even at a constant speed.
It might
be a method for solving
these difficulties to transform the
weves into same
rectangular onesby means of
the ordinary procedurethrough cutting
off and amplifying
the patterns,
and
then amplify itspower accurately by
using a
2o
tilt,
constant voltage regulator, finally
connecting to an ammeter
or a
recorder after
integration andÇattenin
of the patterns.
However
this method may involve unfavorable
points such as variations
in
characteristics of valves,
indefinite accuracy of a constant voltaeregulator and
extraordinarily complicated
mechanism of the
app-ratus.
In this situation,
use of a newly-devised transformer
of
which the core is made from
a special alloy of high degree
of satu-.
ration seemed to have solved all these problems.
Because of its
ex-cessively high saturation of
flux, the current induced in the
secondary coil becomes always
constant regardless of any variation
in form and magnitude of
the current in the primary coil when it
ex-ceeds saturation current of the core.
Accordingly, if theprirnry
current picked up is amplified beyond its saturation
value then the
Induced current in
the secondary coil will
not be influenced by
the
variation in the amplitude of the
vibration or voltage of
power
sup-ply, indicating a value proportional
to the
frequency of the
vibra-tor.
At that time such
a special alloy was first
manufactured by the
Tôhoku-Kinzoku CO.
, and the test resultsverified its
suitable
characteristics for the
desired use. Evenwith a variation in
the
voltage of power supply of 60 to
120 V, for instance,
nothing was
found to be wrong in
the test results on the transformer ofthis type
(loo V standard).
This special alloy
was soon put on the
market
under the name of "cen-delta".
The first set-up manufactured for trial is of
a rectangular-waveconverting type with a constant-voltage regulator, and the second one
is of the special transformer
one was also manufactured ror trial, so a detailed explanation
about the first two set-ups will be omitted in this
article,VI-2 Details of the indicator and the
recording apparatus
As has been discussed in the previous article
the fundamental
problems are ali
settled, but stillremains
the following defectsin the second set-up
None of the ammeter of a simple and compact
construction and
yet of a high accuracy is available.
It is impossible to measure the
characteristics of sea
waves
because of its large time constant.
) The calibration
system is imperfect
due to its mechanicalfunction.
As the set-up contains many valves, a reduction of the weight
and
electric
powerconsumption
cn
hardly be
expected.
It sometimes gets
out of order.At the time when
the second set-up
was designed the main object
was to measure only a
ship speed, and therefore
the time
constant in
the circuit of the set-up was taken as
large as possible in order
to
avoid unnecessary
movement of a needle in the ammeter due toan
instantaneous effect of
sea waves.
However, as will be seen fromVIII-3, it has become clear that the speed meter
is also useful for
8n
observation aI' the characteristics of seawaves, e.g. wave height
and its length.
From this view point the
recording apparatus would preferably be
operated at either the same speed as that of the original vibration
or a reduced speed,
the tormer
method being
used in the
cese of
mile-post test
or observation
of the
sea wave
characteristics
and
wind effects,
etc., and the
latter for
a measurement
of the ship
speed and
travel distance.
Taking all
these into
consíder*tion the
third set-up
was manufactured
for trial and
its mechanism is shown
by a block diagram in Fig.
6.
This may be
interpreted in the
followings:
In-put to the set-up Is
fully amplified
at firat and
then
led to the
primary coil
of the
special
transformer,
induc1n
a
current in the
secondary coil exactly
proportional to
the frequency
of vortices.
The induced
current is
then carried
to en
ammeter
after being
integrated and
also made
flat through an integrating
e-quipìnent.
The ammeter
;ives a direct
reading of the
ship speed.
This part of the
second set-up
is also the
same.
The ammeter
should include
a calibration
system.
It is con
sisted of
a standard
wave of 50
cycles per
second that hes been
di
tulnished
successively
from a 400
cycle wave
caused by
an electric
oscillator.
Calibration
can be done,
if
necessary, by operating a
8WitOh
introducing the
standard wave into the
amplifier in
place of
the in-put.
Rotation of
a pulse
motor driven
by an
amplified power of the
in-put (
the frequency
of which has been
gradually
decreased
)
can
be transmitted
to the
recorder after
being reduced
to a suitable
ratio by
assorted gears. This recoider
cn be used for obt8ining
mean speed
or a travelling
distance of
a ship during a long
period0
The recorder
is also
useful for
recording the frequency of
en in-put through its direct amplification.
This enables us to
see a speed variation within
short time, end thereby
this re
cording system can be utilized In
ship tr11s or in a case of
ob-taming the characteristics of
sea waves.
The two sets ot
empli-fier illustrated in the lower
part of Fig.6 are to be used
when a
ship-equiped type speed meter is calibrated by a towing-type
one.
The detailed circuit diagrams
are illustrated in Figs. 7 and
8,
and Photos, 2,3 and 4
show the apparatus of
a test product.
24
5)
In expectation of an accuracy
for the time signal the above
mentioned electric oscillator is also used to send 400 cycle
waves
which are recuced to 1/8 of the original frequency and then led to a
pulse motor after being amplified In
power.
The rotation of the
motor is again reducth to
a suitable ratio sending the time
signal
to the recorder.
Providing against its
emergency a changeover
switch is readily attached for
the time signal of outside
or an
1 ¿ -' _a:4I_j .IIl!: I I1 I t I
The cable used for the first time was of a four circuit cabtire
cable having tin-coated steel wire (as tension member) in its core.
Currently used is a single circuit type cabtire cable which is
corn-posed of insulated wires of tin-coated steel.
The details of the
cable are given In Table 2.
VIII TEST RESULTS OF THE
SPEED METER
During a stage of the investigation on the trial manufacturing
of the speed. meter many tests have been conducted on the actual
ships at sea, and. Borne of the Interesting results will be described
in the following.
VIII-1
Distaflce of the preliminary run at a mile post speed trial
On the occasion of the mile-post speed trials, the speed variation
at the preliminary run, which takes place after a completion of
turning and settling the rudder amidships until she reaches a mile-post course, has been measured and recorded on not less than
ten cases by means of the newly-designed speed meter. The principal
particulars of these ships and the data necessary for the analysis
are shown in Tables 3 and Li
respectively.
Relationship between the speed recovery and the time elapsed are illustrated in Figs.9 to 15. From these resulte it can be observed, in general, that the distance of the preliminary run before entering
the mile post course is not sufficient enough to reach a constant
speed,
and. this is especially pronounced in the case of
large sizedtankers To find out the distanee of the preliminary run required
for a speed trial, the following theoretical some analysis is made on the basis of the test results.
The equation of motion and its solution are given by,
(1)
= T
V; zs =
'-w-
/
0e I T?"
26
-where
Vo
ship Bpeed when therudder le eettled. amidships
after a completion of turning,
VI =
terminal speed of ship,
Tj
thrust of a propeller corresponding to the
terminal speed,
w i::::
displacement of a ship,
si
distance of the preliminary nm
from V,to V,
k =
coefficient of the added mass,
T
thrust of a propeller to maintain the speed y.
According to the test results on the Greyhound, the value. of k
was found to be 0.2, however, Lamb and other latest exerimental
results give between 0.02 to 0.07 for ships of block coefficient
virying fro'm O4 to 0.9, and therefore it was taken as 0.05 for each
ship.
At the instance when a ship enters into a straight nm
after n
completion of turning, revolutions of the propeller within an unit
time are still lower than its terminal one but on the other hand its
slip ratio is higher than usual, and therefore the thnst
can be
assumed to remain constant T
As can be seen from Equation (11), S
becomes infinity when V is
taken as equal to V1
,and thus the value of 5. corresponding to
V = o.q7vj
was assumed to be the neces8ary distance for the
pre-liminary nin.
When computing s,
, it becomes necessary to know the value of T,and therefore the thrust constant
Kt= T/ çncfr
the advance constant of the propeller
3 = V1 'n[)
corresponding to
28Assuming that J is equal to 0.45 for any kind of propellers, the
Kt values were determined from the Design Chart of A4-40 Type of
T.T.R.I. for propellers with four blades and
from that of ATJ5-65Type for five-bladed propellers.
(See Table 5 ).
The calculated values of S based upon
the foregoing assumptions
together
with the measured Sare shown in Table
6; furthermore acomparison of their results in terms of a non-dimensional form,
1/7Y3
is shown in Fig. 16.
From
these
results it can be easilyseen that the calculated values are generally
in good accordancewith the
actually measured ones In spite of the assumptions made forthe computation.
It is also found that the variation in
the re-.quired preliminary run due to the
difference In theout-put of main
engine is relatively small, being less than about 13%. Thus using
the results for the 4/4 out-put of the
main engine
therequired
distance
of the preliminary run can be determined from Equation (11)or
from the following analytical relations,--
160
ç7Y3
for an ordinary cargo ship,
JyZ70
for a large sized oil tanker.It should be emphasized that.a comparison of
tank test resultswith
those of speed trialsof an
insufficient preliminary run is notonly useless but also liable to lead to misunderstanding for the
sh11ow water effects. This is especially
important
in the case oflarge sized oil tankers for which a very long preliminary run is
required before entering a mile post course.
VIII-2 Error in the milepost distance and the shallow water
effects
It Was reported that the propulsive performances under service
condition of the ships built at a certain ship yard were always
significantly better compared with their trial results, and the reason of this phenomena has been believed as the effect of the
shallow water in the
mile-post
course.The depth of the water in the nille post of 19 to 22 meters as
given by a chart is shallow indeed, but since there is some doubt
oÍ a possibleerror in the mile-post distance, tests were conducted in both a mile-post course and the open sea by using the newly
de-signed speed meter.
The test results are shown in Fig. 1? where the speed variation
against time is illustrated ( the data marked i to 8 are for the
mile-post results, whereas 9 to 12 at open sea ). The mean speed of
each return trip as compared with the results of
ordinary
mile-posttests
(measured by the
shipyard ) are shown in Table 7 and in Flg.18.As can be seen from Fig. 18, it has become clear that the method
of the mile-post course speed trial may involve not
only the
shallow
water effect but also an
unexpected error in
the mile post distance.VIII-3 easurement of sea waves
Since there would be no possibility of existing a time lag effect in the mechanism of our speed meter, the indicated speed should
in-dude the effects of both the ship ïnaking weves and sea waves.
Ac-cordinly it will be possible to obtain also the chaxacteristj.cs of the sea waves, 1f necessary, by analyzing speed curves recorded by
the speed meter.
With this object in view,
the followingseveral tests were
carried out on the LI.S. "Argentina iiaru" on the occasion of her
re-turn voyage
from Yokohama to Osaka.
VIII-3-1 Waves made by the ship
A) Wavelength
The cable of the speed meter was drawn behind the ship with
a length of 50 meters, and then at every
5 meter
interval itheld for 30 to 50 seconds to take readings for
the
measurement untilits
length
reached 400 meters. The recording of the results was done by a method of penwritlng of the originel vibration.Since
the frequency of the sea waves, in
general, is considerablylower than the above-mentioned period,
the effects of sea waves willnever appear in the
meanspeed of the
recordedresults which include
only the
effects of the wave made by the ship caused by itsmolecu-lar motion.
The recorded speed along
the distance from the aftperpendicular of a ship are
shown in
Fig. 19. The heavy solidline
in the figure indicates
the ships speed which is varied with achange
of such conditions as the course direction, wind force, etc. ( For
the
detail, see
Ref. 16 "Notes on the TJse Ofa Towing Type Speed
Meter and its
Application
to the easurementof Sea Waves" ).
As30
J
-f-
-
e2Trf
2 2
( Z)
can be seen from Fig. 19, the speed Indicated by
the speed metershows a periodical change with a variation
In
the cablelength,
while the relative speed of the ship against water (
the heavy
solid line
)indicates a
monotonous variation. The maximum or theminimum speed as appeared periodically in the figure means a
rela-tive speed when the speed meter is
locatedat
the troughor the
crest of the wave made by the ship, and therefore the
distancebe-tween
the two
adjacenttroughs
( orcrests
) in thefigure
indi-cates
nothing butthe wave
length made by the. ship. Thisfact
canbe clearly
understoodfrom the results shown in Table 8, indicatIng
a close
agreement between the theoretically obtained wave speed andthe
mean
value measured by the speedmeter.
Thus
itbecomes
possi-ble to
determinethe wave length by
using the newly designedspeed
meter of towing
type.B)' Vave height
As the speed meter of
towingtype works underneath
the freesurface of water, the
indicated velue correspondsto a
speed at itsdepth ( d meter
) Instead of showing
the true
speed onthe free
surface.
The correction should be made by
the following forfllula,where Vmax. end Vmin. are the maximum and the
minimum speed as
given by the speed
variation curve in Fig.19,
wave length, and
the suffixes o and d of the
brackets denote for
values at the freesurface and at depth d, respectively.
Then the wave height at the
free surface can be given by
H0 = Z
± (V1 - Vmin)o
2-
(f3)
The wave height at
a position of 2.3L behind the aftperpendicu-lar of the ship running at speed
of 18 knots is calculated
as shown
In Table 9,
where d is assumed to be 5
meters, no actual databeine
available.
A theoretical value of the
waveheight
for atwo-dimensional
model with similar principal
particulars to the M.S. "Argentina)eru"
Csee
the later explanations) is also given In the saine
table, and it coincides fairly
we1l with that calculated from the
actual data.
It
can therefore beestimated that the height of
the
wave made by the "Argentina Maru" at speed of 18 knots was
approxi-mately equal
to 50 cm.
VIII-3-2 Sea waves
The above-described
Investigations have fullyverified that our
newly designed
speed meter is ofa quite satisfactory sensibility
end
accuracy for detecting the
effect of waves.When the speed meter
is used in sea waves it should be able
to
recordtheir
effects
in-elusively, and thereby to determine
their characteristics from
an
a-nalysis or the records
With the object to check
up on the matter,
the following theoretical
study will be made.
Using the notations
2
-T¡g
'b
4 I27Í . 2\_T
9t0L+v'
H =
I 21t9 y' ,k4fcosQe ]
wave òhead
Ö<<oD)
The value of ( Vrnax-Vmin )d is the difference between the
maxi-mum and the minimaxi-mum
speeds
indicated by thespeed
meterwhen a sea
wave passes over it and is obtainable from the records.
Since thetime required for a sea wave to pass
over
the speed meter isgener-ally very short, lt is necessary to get a detailed aspect of
speedvariations
dueto sea waves when the length of the cable is kept
constant.
. ,II'tIIl ¡ ti
1
L
(1+)
4/i
:i::VÖiow:nwdVe (Ç6<8;)
¿ cogo(
X
(Is)
T
period of encounter,
oK entrance an1e of sea
waves with the advance
direction of a ship,
V :;:
sh1p
speed,
41J:
:::
wave length ofsea
waves,
4.
Ho
wave height of sea
waves,
vo ::::
advance speed of sea
waves,
the length and height of
sea waves can beexpressed as follows;
Three examples of speed
variations
for O to 50 seconds were chosen from the records of test results on the "Argentina Maru" asshown in Fig. 20. Each plotted speed in the figure is the mean
speed corresponding to ten cycles of vortices. The aspect of the
speed variation due to the sea waves can certainly be perceived in
Fig. 20 except for No. 3. s the condition for No. 2 is
a following
sea, the variation in the mean curve shown by a solid line should be
considered es the one due to the see waves. The calculated values
of H and ?LØ are given in Table lO
As can be seen from Table 10, the length and height of sea waves
obtained analytically seem to be reasonable. It is therefore
con-cluded that our speed meter used as a towing type can be utilized for both the measurement of ships speed and the observation of the
characteristics of sea waves.
Vhen our speed meter is used hereafter for this purpose it must therefore be necessary to find out a method of knowing the actual
depth of the speed meter under the free surface of water..
34
VIII-3-3 Informations
1) As can be seen from Eq. (12) , the smaller the depth of a
speed meter under the free surface is the closer to unit the
coef-cL
ficient 2lT becomes, and thereby the errors accompanied by the
correction becomes smaller. However, the minimum depth should not
be taken smaller than a limiting value for which a cavitation takes
place behind the cylinder. The limiting depth at 20 kt, for
example, is approximately equal to 2 m.
The vertical component of the speed induced by the wave
motion Is neglected. This does not influence on the results of the
calculation of wave height, because the verical component of the
velocity vanishes at both the trough and the crest of the wave.
As have been already mentioned, rate of the speed variations
must be known when calculating the wave height. Since this rate is
influenced by various ship motions, degree of an error due to the
ship motions should be investigated0
If we take the effects of pitching and heaving into consideration,
setting aside those of surging, yawing and rolling, the maximum error will, at worst, be less than that given by the following
formula, where tiV V-
i 1 ( j
j'7r 4 -F-&5iflU),t ,0'
xocrì.L=length of the ship between perpendiculars in meter, k=1/L ( i is the length of the cable exposed above
the free surface. )
=h/L ( h is the height of the supporting point of
the cable from the free surface. ),
=pitchlng amplitude,
w,=angular velocity -xr. fj ( L in meter ).
(I 6) 35
This relation indicates that their effects become large with the
Increase in L, ì'. , and kand with the decrease in k,,
ç = .:
-I -i
36
"Argentina Mani", AV/y
becomes 5 to 6%.
AccordIngly, when an
exact wave height is
to be obtained,it should be calculated arter
rf p
these,due to ship motions are analytically excluded from the
records.
However, these effects
are comparatively small and becomeneglegible if k, 18 taken very small.
VIII-4 Length of the towing cable
As the local disturbances in the vicinity of ship hull caused by
her movement diminish rather soon,,their effects on the speed
measurement may be
avoidable if
a suitably long cable is used, whilethe waves made by
the ship willnever damp so quickly (see Fig. 19)
In order to see the whole picture of this problem, a theoretical
a-nalysis will
be made on
the distribution of theinduced velocity,
u, in the running direction of a two-dimensional ship model
( at
uniform velocity V ) of which principal particulars are similar to
those of the "Argentina Maru". In this analysis, the verical
in-duced velocity will
not be considered because of its insignificantinfluence upon the final results, the maximum error due to
its
Ignorance
being given by
cg
--
- . ç- ) - C7)
It can be seen fr the results of the calculation of the errors
in the
case of
the "Argentina Meru"that
this assumption isquite
satisfactory. ( See Table 11)
Thus neglecting
the vertical induced velocity and the effect ofthe local disturbances, u is expressed as follows,
:C) ' 2
LI - ! (°f°
dd5
L xSeCcos[(Xsecd
( 18 ), .L2 X,sec
IC
j;
whe re
L =: length of ship,
B
breadth of ship,
D
draft,
?L
is a function indicating the ship
form, and can be given by
the following expanded formula In case of a two-dimensional model
symmetrical with respect to its midship;
2. 4
11=_ a0-t- a2i-O4
+ --
(q)
Then Eq. (18) leads to the following approximate tormula;
LL - :
8
lit. L
t-e
L
)
jp%(*I)
.CO5?(1) -
---.. .A'2)
(*i)
(-t)
ççsX('s I)
sin(Ç I)I- ---
[4I)
--- - Jhç-:)
:2X/L,
: g/2V
B7 ---.?/$,
(20)
IÍJ, 7 - '::"A1 I ': ILT II
where
(;) = C()+ D(),
Ç 7 (271ff) - t, (2)))C ( )
,
f(?) = C(') DP(),
The following relations fori are used depending upon each value of
theblock coefficient of ship;
() =
j_
2 4,
for Cb=