THE 1WERLOSS FACR IN SHIP SERVICE PERFOEMANCE
by
Prof. E.V. Telfer, D. Sc., Ph. D.,
Tre ndJae im.The following note has been prompted by a study of Prof. Aertssen's recent I.N.A. paper oi the service performance of the Victory ship TERVAETE.
In this paper Prof. Aertssen showed that propulsive loss broadly was directly proportional to such disturbing influence as wave height, wind. pressure, pitching anglo and so on; and further that the loss was greater the smaller a vessel's relative power. In two papers read by the present writer before the N.E.C. Institution in
1926-27
and in1934-35
these basic facts of the sea behaviour of ships were previously demonstrated, As this earlier work is now capable of further useful development a briefrecapitulation may be excused.
In the
1926-27
paper it was shown that by rearranging a vessel's service statistics into the four distinct weather groups nautically labelled fine, moderate, heavy and very heavy and using the generalised power diagram then developed to determine from the group average speeds and propeller revolutions the corresponding true average indicated horsepower, it became possible to calculate the average admiralty constant2/3
V3/SHP for each weather group. Applying this analysis to the data for a large number of ships it was found that the loss in admiralty constant between successive weather groups was substantially constant. This basic statistical fact at once made possible the develoment of a simple numerical scale of weather intensity, since it implied that loss in admiralty constant was directly proportional to increase in weather intensity, By adopting the scale, zerofor fine weather, 100 for moderate, 200 for heavy and 300 for very heavy, the average weather intensity on agiven voyage was obtained by adding the total
time percentage of moderate weather on the voyage, to twice the corresponding heavy weather percentage, to three times the corresponding very heavy weather percentage. The total f igire placec.i the mean weather correctly between the
constituent classes; and when voyage average admiralty constants were plotted to a 'oase o±' this weather number, a lin relation was usually found.
Vhich, with
= KW/C0
This shows now that the power-loss ratio was directly proportional to weather intensity; and so long as all the constitutent physical disturbances of
weather increase linearly with our weather number our expression I agrees with the first of Prof. Aertssen's findings.
-Such diagrams were generally based upon constant power statistics, but assuming for the moment that the diagram held good for wide ranges of power it can also be interpreted on a constant speed basis, Thus let Po be the shaft horse power required in fine weather to produce the assumed constant speed and. P the corresponding power required for the same speed in weather of intensity W, then the relative power loss is given by
,
2/3
A2/3
v3 1/Po - ip P - p P P o 2/3v3/p = C , can be itten o o,2/3
v3/0
1/po PIf we construct a normal speed and power curve (fig. 2) we see more clearly
the significance of (P_P/
The value (p-p ) is lost power so far as
o
transport is concerned; and. the power-loss ratio is a measure of the Z relative power wasted in overcoming weather.
Dur earlier finding that admiralty constant loss was directly proportional to weather intensity, i.e. that
¿/3
. v3//3
v3/p
= KW reduces to2/3
3P - P
( C = KTV3
3. In between the 1926-27 and. 1934-35 papers a shipping slump had set in.
For this very reason the author in his professional practice carried out a larTe amount of investigation work into the inefficiency of reduced. speed-running of merchant ships; and. inter alia, found to this then consternation that the liner loss in admiralty constant ..'ith weather intensity certainly no longer a:plied, It was boon evident, hovever. that for a given weather intensity the loss was greater the lower the power with which a given weather was faced.; and. that the greatest departure from the previous full-power
relation was always at the heaviest weather intensities. To interpret this experience fig. 1 had. to be developed to the form hon in fig. 3. For each
constant power a separate line was required. All lines radiated. from the same value at zero weather but had. a greater slope (i.e. loss) the lower
the power, Jhen this slope variation was plotted to a base of power for a large number of ships, the plot for each ship IJaSof a clearly hyperbolic natue, a fact which suggested. that the power-slope product would be constant. This was found. to be the case, so that instead of' the admiralty constant loss being solely proportional to weather iatensity it was also found to be
inversely prorortional to the absolute power on the voyage. To use this fact statistically the somewaht artificial concept of weather per 1000 IHF was introduced.; and this as once stabilisod and unicuely linearised the voyage statistics.
We now see that instead of' the expression 1 above, our expression ':ith variable power running shows that the relation
(r-r)/p = K1TT/CP 2
is required., In principle, this again agrees with Prof. Aertssen's firflings, It is evident, however, that this expression still further simplifies to
(pp0)
= 3In other words we see that for a given ship meeting constant weather the power loss is ind.ependant of the absolute power. It thus follows that all iso-weather
(isonduir T) power-speed curves are naturally paralle' to the basic zero weather relation, implying therefore that the speed-loss in given weather is much less at high power than it is at low, a wellknovi fact in ship experience.
which L is ship length in feet. This expression being dimensionless obviously reruires the weather intensity also to be dimensionless. If weather were
measured by wave-height in relation to the length of ship, which could for xample be an acceptable definition for model experiment purposes, then we could say that
P-p o = K . (H/L) or calling (p-r), \ P we have = Constant H
in which H is the wave-height in feet, This relation can be conveniently called the power-less factor due to waves. It obviously would be extremely useful in reducing to a single value the results of model tests over a range of ship speed (clear of resonance, of course) and. over a rango of wave height. If model
geosimos were tested one might reasonably expect the power-loss factor to be independent of scale. By varying the wave-length for a given height the
influence of A /L can be examined, The behaviour of the factor over a
sufficient ). /L range would supply an acceptable criterion for one of the
important weatherly cualities of a ship form. For our present purpose it is sufficient to state that model experiments do endorse the power-loss factor concept.
5. When weather intensity is measured only by relative wave height the idea of a dimensionless weather intensity is simple to comprehend. When the weather, however, is descriptively defined but numerically interpolated as in
the authors system, the physics behind the statistical treatment is not so - 4
§ 4.
How can we best study the implications of this power-loss constancy ? ilst in our earlier work we used the admiralty constant, we could eçually well use other pseudo-dimensionless presentations. As the power-loss isvidently independant of speed it is probably simpler to adopt a dimensionless presentation which does not involve speeds A suitable form would appear to be
(p-r)
w
\,
5
obvious. Provided, however, that the ships! officers give a true description of the weather and sea conditions and one which is not subconciously influenced by the size of ship which they are in, then the resulting weather intensity may be regarded as directly proportional to some equivalent wave ight : and
expression 5 could be retained for stastial analysis in the form
= Constant 6
and correlation with the model experiment in waves rould determine the eçuiva-lent wave height in terms of weather intensity. Actually, moderate weather
appears to correlate with a wave height of 6 to 7 feet and the length dimension
of weather may therefore not be too unacceptable.
e can call (6) the statistical power loss factor. On any particular voyage. the actual average power can be determined by the author's generalised power diagram. The zero weather power corresponding to the average speed can be derived from a suitable analysis of load trial performance extended to embrace a range of draught. The power loss is thus known; and hence the statistical powerloss factor can be calculated from the known weather on the voyage.
Actually for numerical convenience, the statistical powerloss factor is best
expressed by
P
= Constant 1000
A
This factor can be presumably applied to all ships independent of size and will be a criterion of weathert-1 merit, independ.ent of poiiTer, size and weather
intensity. For a given ship the steady increase in the value of the factor with time out of dry dock, or its rapid increase on actual fouling of the hull,
should serve elso as an excellent indicator of hull surface condition.
It is submitted that the use of these two powerloss factors, the wave
powerloss for models and -the euivalent wave or statistical power loss factor
for ships, should prove extremely useful research instruments, assisting the better understanding of the sea behaviour of ships; and hence contributing ultimately to their better design and performance
400
30e 200 100 o o F M f00200
300
WEATHER INTENSITY
FIG. I.
o VH O lOO200
3oo
WEATHER INTENSITY
F M HFIG. 5.
SPEEDFiG. 2.
H VHi
PI
w o Q-cl,o
400
3ooAC.
200 tooAFXS (I
I1s P/îi CQPAiSON
by
I. J. P.ALLAN, Superinterzient, Ship 1)ivision,
atiorl ¡ysicsl Labors tory.
-= - -== =-= -= .. - -=-=-=-= - -
--The subject of this Syziiposium is "Trials of ships at Ses.
This is
taken to refer prLrnarily to Measured Uile Trials, with goî coriittions of wirxl,
weather az
ship, an-i also to service porforuioe so far as that can be assess ed.
Airoest all ship deai
today are based on the resulte of nodel teats
both for resistance ani propulsion.
This practice infers tIt the recuits of
a1el tests can be interpreted accurately on the full scale both jualitatiely
azzi quantitativoly, arzi it wi].1 be of ntcrest to cousider in seine detail the
fact ors
nvoIved.
The coerparison of the resulta of tests on ship models with the results
of full scale trials is one receiving particular attention in recent
years, chiefly
because of a geneml desire to sxplore azz3 urderstaM tho "factors of iiorance",
azi. also because of an increasing appreciation of the eff eat
on perforaanoe of
hull roughness, both structural arzi due to fouling.
This is true both on a
national ani en international basis,
Special research on this subject is
certainly going on in U.S.A., U.., Ibilani azxì Sweden, sui
we have the work
recently carried out by the Centre kielge do Recherches navales
on the "Torva ate"
sl ro1xsed for another vessel.
In a'ì-ìtion, the matter has been cLiscussed
from various angles at the International Conferences of Ship Tank Superinterdenta.
Letailed results of various investigations in hazii have not yet been published
so full discussion is z,t possible at this stage, but the various aspects
can be revi awed.
Th. ship owner suite rightly considers the perforìenoc of nia ship
in service sa the real couercia1 criterion, but it is difficult
even on a
statistical analysis basis to asacas service perfonnce
accurately.
The
increase for
average
service perforsnce over
seured mtl. p.rformenoe varies
from some l5
to 2O
on the iost favourable m* routes to some 1
to
451on the least favourable routes.
In view of that has been sd above euch
percentages cannot be other than very rough arzi it is ccnclz1&
that the ship/
eadel oorçarison should in thc first place be based
on a measured nile
perfonce.
thile accepting that position for basic OOflFi$E
cene should
not lose sight of the importance of
desipp for scaIiiilinevs, a quality which
is not tested on the eured
1 e trial axi which has not rec4vi sufficient
attention in the past.
The probln of correlating the nodel
resu1t$ to the ship result
was first suocoesfully solved by the Frc*x3e
method in the 1rtter pert of last
centu2.
The basic principles of that method continue in use today, although
some departures in detail have been
introduced.
Gne of these departures is
the universal adoption of self-propelled
tests in place of the earlier tests
i'vith propellers supported from
frames behiz
the hull model.
Another is the
use Of a different method to extrapolate
the model resistance result to the
ship resistance result, i.e. the
surface friction correction.
It may clafify
thoughts if we consider that there
are two
approaches to this
tter (1) from the designer's point of
view, and (2) from
the research worker's point of viow.
AS regards (1) we re only concerned
with a rcaonabie prediction of the ship's performance, and so long as s useful
set of correlation factora has been ,orIced out from xperiere it is not
iportont
which basis of extra>olation is used.
In other words so long as the
extrapolation
method was reasonably in line with
the physical faots, the correlation
factors
would look after the rest.
The method is effective in dealing
with normal
designs but does not give such oonfidce whc one is faced with unusual designs.
It should probably be adiitted that ifl the pest research workers have to a
o.aidereb1e extent been content with
t his approach.
As regards (2) it is
important to ezilain and assess in detail the various parte making up the
reels tance aM propulsion
picture both of the model and of
the ship.
igeinst
this bnckground the control
of flow conditions on the modele the accuracy of the
metha a of extrapolation of resistance and propulsion, and the correct
asseesuent
of the ship performance melding
knowledge of the roughzims effects
on the ship,
ail assume
eat isportance.
The interpretation of results end
the ultiate
value of the deductions
depes on the success achieved in these directions.
Taking the
del ed first it will probably
not be diauted tha.t the
resistance (including speed) is
known to within
T1 and the propulsive efficiency
to within 2.
The hull analysis factors
are not de ¡irìed with the same
accuracy.
It isay be assumed that the xx1el hull and propeller aro technically
5(500th, i.e. they áo not
oase in frictional resistance due to
roughness effects.
It is isportant
)
I1&U]that the bouniar7 layer flow
is t'bulent, aM this is g emily a
bievecì by fitting suitable
stilatore
at or
ar the boy,
th
al.
so far as the model
propeller is coricerz
zio special preceutiozie are
taken except to evoii the use
of very sii%dl propellers
(roto critoriozi tìuoptet by I.C.T..)s
It is assumed that flow coiiti
on the
propeller are turbulent.
Root researches izïiicate that this may
not be an
6itirely
tisfactozy assumption, although the effect on force measureenta of
a limited
eiflOUXlt 0f laminar flow ta prcbably within the accuracy
of the experiments.
If the normal 1rode extrapolation
coefficients are coepard with,
sr, the ckoenherr
line for eaooth turbulent flow, it will be founì
that the
ente give
exoOsE over the
ohoenherr lino which varies idth
the size nd apev
of the vessel, the excess being of the oror
of lQ
for
horter 1engtia
v1f GD log!er length.
It is i,lear, therefe, that the
Ñ'oe CfjL
:iits
.l1 lot L4v
consistent rult for all lengths az1 speeds
of vesse1a.'
f ri0t1i liae
the above dim
rwi at low noi4
sae evjj erxe to show that the true
inimnwn turbulent
h,jbe Ateeper than the 5choeztherr
et.nd if that is true
4'
11 be j flcrcssed.
The epecif
of
odel
sp
3d13 Ue above the
herr 1ir,
alzi This incresc
zwy be dcribed as fm resistance arising partly froa
VelocJty e1f0t
on tho
friction azI partly frocs unbalanced preseure effects.
The question of the
crrect way of dealing with this "fozei effect" arr! whether or not it is subject
to søale effoet is a matter of importance which is being investigated
at the
present ti.
Other methods of overecnirig these difficulties he#e been
proposed
but they will not be discussed here.
It is not general practice at the mOEsont to make
any adjustment of
ax,del pr.dietions for scale effects in propeller thrust azzl torque,
or in
ke
factor' aM thrust deduction factor.
These are known to exist in varying
degrees aM until satisThctoxr methods of allowing for thes
are worked ont
the dtsiled comparison of ship sod model results reterru to earlier cannot
be achieved.
As regarrs prptller faotor
tie cre i.ti
cts
e svoidul by
arznging the scale BO tht a oerte.tn rnixLRv.
ria eeed. on the blades
but
there is still a good deal of iwrsrLce regardthg slow
seals effects which xY
take 1ace between this point aM the full scale.
Efforts
re biflg je to
explore this field.
tts wake azrl thrust dIuatiorL scale
etfect it is probable
that the 1&tter ja negUgible, but it is krovn
that the f oziner is
sterial1
w
the writer has propos
a
ethcd of dealisg with this in a z'sct paper
to
the North East Coast Institution of flgineers &
thipbuilders.
There is always
an e1cent of doubt reerding the
effect of ship's roughness whith
texas to
offset the reduction of wake factor on th. ship vs
coçred with the sod el.
The application of this iske correction to a abcr of good trial results
iïLtoat* a good areenimt between torque coefficients on ship aM
del but
a considerable eesa of ship thrust
coefficient canpared with the rodel.
In gere1, t1efore, frc
the nodel
mi, *hile the baLto rults
are known to a good degree of aceursay,
the abolute value of the ship preijotions
is
xzr in
ø doubt for the varis reesons stated.
Turnuzg now to the ship sido of the picture, it
is beccxning increasingly
clear that there are very nany difficulties in
obtaining ccpletely satisfactory
ahip trial results.
Ordinary eoercial trials are subject to a number
of
possible errors, and the cxIparison of the results
with the predictions from the
model teste is to that extt fortuitous.
it is important that the ship surface is snooth a:i
clean.
Experience
S LIII'?
has shn that a surface which looks clean but
has a
feeling can increase
the pcy,ior to a material extent.
ilecent research has also shown the important
influence of structurel roughness ori resistance so
that anart frc
the saving
S