Vous êtes sur la page 1sur 10

Skin Friction Resistance of Ships

....

and the Problem of

By Dr. F. H. T o d d J

Extrapolation from Model to Ship


gave frivtiomd coefficients for different lengths of pl~mk.
This e n t i r e technique, as exemplified by Equalion
[1], may he c.dled the "Froude assumption." It ha~
stood the test of time so well that. it is still used by all experiment tanks and, however imperfect, still remains the
basis of all methods of ship power predictions from the
l'esuJis of model t.ests. The two methods at pre.sen~
agreed upon by the International Towing Tank Conference for use in all published work are based on this assumption, one using the Froude friction eoeffieients, the
other the American Towing Tank Conference 1947 If,e,
originally derived by Schoenherr.

T h e International T o w i n g Tank Conference


( I T T C ) is to h o l d its 8th m e e t i n g in Madrid in
September of this year. One of the subjects to
be discussed w i l l be the perennial one of h o w to
estimate the resistance of a ship from that measured o n a small-scale m o d e l in a t o w i n g tank.
T h e Skin Friction C o m m i t t e e of the Conference
w a s charged at the last m e e t i n g in Scandinavia
in 1954, with r e v i e w i n g the available data and
m a k i n g r e c o m m e n d a t i o n s to the Conference in
Madrid w h i c h will, it is hoped, be universally
acceptable. Such a d e c i s i o n w o u l d r e m o v e one
of the principal difficulties e~:perienced in the
use of m o d e l data in comparative studies. It is
believed that a r e v i e w of the present status of our
k n o w l e d g e in this field may be of interest to the
m e m b e r s o f the Society at this time.

A Common Formulation?

The proper correlation of the resistance of ship models


of different sizes and the reliahle prediction of actual
ship resistanee from model results lie at. the very foundation of towing-tank work. The first step towards the
solution of these problems was taken by Williain Froude
when he proposed to divide the total model resistance
into two parts, one due to wave-making, which he called
residual T resistanee, trod the other due to frMirmal resistance, or, in coefficient form
(',1 = ( ' , +

(:~ . . . . . . . . . . . . . . . .

{1]

He showed that the wavemaking resistan('e would


scale directly as the displaeement at "corresponding
speeds"; i.e., at the same value of V/X/L, ('~ would be
the ~-ame for any size of geometrically similar hull. To
determine the frictional dr-lg he made the assumption
that the frictional resistance of the curved surface of the
model or ship would be the same as that of a smooth
plank having the same length and wetted surface, and
Technical Director, tlydr.meehanics L'dJoratory, David Taylor Model Basin, Navy l)eparimenl, Washington, I). C. Member
SNAME,
JULY, 1957

What of our present knowledge of the subject, and the


chances of agreement on a common formulation'?
in 1!154 tile Skill Friction Committee of the I T T C in
its report stated that it believed a proper correlation of
model and ship resistance must take account of the effects of lhree-dimensional flow, and that a suitable basis
from which to develop a method for such eorrelation
would be the smooth turbulent friction line for two-dimensional flow. The experiments considered by Froude,
Schoenherr and others were made on planks of a variety
of sizes and shapes, geometrieal similarity being ignored.
The result.~ suffered in varying degree ft'om lack of turbulence stimulation, aspeet ratio and edge effects, and surf~we wavemaking.
Ill the past few years Hughes has published results of
careful experiments on planks and pontoons covering a
great range of size, and by extrapolation has derived a
two-dimensional line for turbulent-flow friction over a
smooth surface (1). " This is eompared with the American Towing Tank Conference (ATTC) 1947 line in Fig.
1. The Hughes line is higher and steeper for values of
log 11 below 5.65 and lower for all values of It above this
point, although in the ship region it is somewhat, le,s
steel) than the ATTC line. It should be remembered
that the llughes two-dimensional flow lille is derived
from three-dimensional data and so depends upon the
method of extrapolation used. That it is not necessarily
a unique line may be concluded from the fact that
Weighardt, using Hughes' data hut a different method of
e Numbers in p-trentheses refer to the Bibliography at the end
of the. paper.
3

.,i,4
l

'--T

0004

.,

"

ii

]
,

!
~

'

CURVE

AUTHOR

--(,<s,

,o~

c,

,~ ,,u

r,

l~j , z o

I T T C

COMMITTKE

IT

, ,rMMI'TE

T ~

"%

-1

~
v

Problem

However, the sem'eh for a two-dimensiolml flow line is


only part of our problem. Indeed, there are some who
do not believe that this line is an essential preliminary to
the extrapolation problem, but rather that. each hull
form requires a separate extrapolator. Telfer has advocated this approach since 1927, believing that the specific
4

:,

'

I.L,i

i
1

'

o00l

. . . .

s ~

'-I

I0

Comparison of Hughes' turbulent-flow friction curve with the ATCC line and results
of others

extrapolation, has derived a different two-dimensional


line.
The frictional coefficients represented hy the Froude,
Schoenherr, Hughes, and similar lines are derived from
measurements of total drag. Another method of obtaining a friction formulation is by measuring the velocity distribution in the boundary layer, so determining
the shearing force at the surface and then the resistance
by suhsequent integration. Using this met,hod, Landweber has given result, s which, over the pracJ, ical range,
coincide almost exaet, ly with the ATTC 1!)47 line (2).
Coles, on the other hand, found a line much lower than
the ATTC at low Reynolds numbers, but with much less
slope, so that it almost coincided with the ATTC line at.
the highest ship values of Reynolds nmn|)er (3). Townsend, using a theoretical approach, proposed a line about
midway between Hughes and ATTC, and having less
slope than the latter (4). Some of these lines are shown
in Fig. 1.
AI, the moment all that can be said is that
there is no absolute agreement between any of these
lines, the slopes differing to an appreciable extent, and
the maxinum differen(.e in ('~.. at R = 10"' being some 12
per cent.
The Extrapolation

'

:'

. . . . . .
LOG io

..

:,

i ?b,

! i

"~,,"-4,

. . . .

'
6

Fig. 1

~,

'a ~ g , :

.I

,,

00OI

.%

~r-

. . . .

]
FORMULA

frictional resistance for geometrieally similar forms (geesims) is a linear function of R--'A (5). Troost and Lat~
have puhlished a proposed met.hod which includes a coefficient A dependent upon the hull form in any ~iven
case (6).
It is evident, of course, that the Froude assumption is
only a first approach to the extrapolation problem. The
frictiomd resistance of the avtual curved hull surface
cannot, in general, be equal 1o that of the equivalenl
plank, for the local velocities of the flow are different and
the paths traversed by the streamlines are longer. Wigley has made estimates of the increase in skin-friction resistance on some mathematival forms from this cause,
and has found that it, could amount, to some 7 per (.enl~
even on fine, slender models (7). This increase in resistance, being frictional in nat tire, lnight be expeet.ed to
scale with Reynolds mm~ber at. a similar rate to 1,he
"plank" skin friction. Also, the transverse curvature of
a ship form has an effect upon resistance. As shown by
Landwet)er, the increase depends upon the absolute curvature, and is therefore greater on a small model than on a
large model of exactly similar shape (8). In addition,
some resistance is present due l,o eddymaking and flow
separation, and this may be of significant amount on a
full model. The extent of separation on the model will
depend on the state of turbulence in the water as well as
on the size of model and its shape. The eddy-making
resist.anee itself will scale with Froude number and not
with Reynolds.
The wavemaking resistance is essentially a high Froude
mtmber or speed effect, and at low speeds with most
JOURNAL

O F SHIP RESEARCH

...

(2T

~T =

~ ~ g

Fig. 2

C o m p o n e n t s of C,

models becomes negligible. Thus in earefully conducted


experiments, where care is taken with regard to turbulence stimulation, steady speed and good surface finish, a
point A, Fig. 2, usually can be found below which the
curve of Cr is "sensibly" parallel to the two-dimensional
friction line. This is called by Hughes the "low Froude
number run-in" point.
We can take our analysis a step further now by writin~

(.',,

= c~, +

(',. +

c. .............

[27

~vhoI'('
('~., = resistance of equivale)H plank in two-dimensional flow
(:v = form drag due to shape ()f hull
(~u- = wavemaking resistance
As will be obvious, Cv (see Fig. 2) is made up of c()ml)onents due to additional skin friction caused by cur\")lure ettects, both in the fore-and-aft and transverse (lir(,ctions, to separation of flow and to eddymaking.
These components can)tot be separated in any clear-cut
way, at, least with our present knowledge, since the separation point may change with speed, and the frietional
and wavemaking resistances affect one another '~s speed
rod so wax.(; fo)'inali(m ('hal)g('.
We are thus led lo reMize what a complicated pr()l)h,m this model-ship correlation really presents. ()ur
immediate task is to derive a nmthod for scMing the
three-dimensional viscous form (h'ag represented by (y
in l:ig. '2.
The Froude assumption means, in effect, lhat we
scale all form resistance represented by C'v with wavemaking resistance C')~,, t,hereby assuming that (('v + ('.-)
JULY, 1957

tot,4

= CR, remains unaltered in passing front model to


ship at the same value of the Froude number V/%/~.
In xdew of the fact that some of the resistance represented by Cr is undoubtedly of a skin-friction nature
sealing Mth Reynolds number, this assumption cannot
be correct, however convenient it may prove to be in
practice.
The other extreme would be to some Cv directly with
Reynolds number, ou the assumption that Cv is a conslant percentage of C~. for any given form, leaving only
C~- to scale with Froude number. In effect this has been
proposed by Hughes in association with his new two-dimensional flow friction line, Fig. 3. The "form factor"
Cv/'C~, is determined from the "runqn" of the Cr curve
for the model at low Froude numbers, and is assumed to
apply directly to the ship. The value of the form factor
is a function of the model shape, fullness, length-to-draft
ratio, and so on.
Hughes' assumption goes to the opposite extreme from
l"roude's, and it is evident that, the truth lies somewhere
between them, but there is not sufficient knowledge or
data available ~oday to take the analysis any further. It
is probably true to say that most experiment tanks would
desire an extended period of trial of this method before
they would be willing to endorse it for universal adoption.
Roughness of Hull Factor

When in the future we eventually reach the state


where we can with certainw predict the resistance of
troy model or ship from the results of experiments on another geometrieMly similar model, we shall still have a
further problem. The ship results so predieted will be
for a "smooth" ship, and the necessary allowances must
5

o.oa~

dT- ~eSv~
O ,oo3

O , (mT.

r2 = v__~
L
Fig. 3

Derivation of ~ C / f o r French minesweeper Aldebaran using Hughes' method

be m a d e for the "roughness" of t h e actual hull surface. This roughness is p a r t l y due to structurM roughhess, caused b y p l a t e edges, rivet points, welds and shell
fittings, p a r t l y to paint-surface characteristics and, for
ships in service, to m a r i n e fouling. This roughness allowance expressed in coefficient forin is actu-flly given
the symbol AC,,.
Knowledge of ACv can be derived in a n u m b e r of ways.
T h e results of experiments on different p a i n t s on a 21-ft
a l u m i n u m plm~k run at, T a y l o r Model Basin are shown
in Fig. 4. In general the p a i n t roughness gives rise to a
constant addition to the smooth C~. curve r a t h e r t h a n to
a constant t o t a l value of ('~., as found in sand roughness.
This same a p p r o x i m a t e ('onstancy of AC,., also has been
found in m a n y full-scale trial results of U. S. N a v y surface ships, ACt. v a r y i n g between 4-0.0001 and 4-0.0010,
or a b o u t the same range as found in t h e pltmk tests (9).
Full-Scale Tests of Extrapolation Methods

Values . f ~(/,, also can be derived from full-s(.ale l rial


results, and this indeed is the final test of our extrapolation methods. T h e ideal w a y of obtaining these fullscale values of ",('~,, would be t,o measure the actual resistam'e of the ship. There are m a n y intrinsic difticullies in doing so, and the most su('eessfifl trials of this kind
1o d a t e are t m d o u b t e d l y I hose of t h e L~cy A shiest carried
out b y the British Shipbuilding Research Association in
which the vessel was propelled b y air('raft jet engines
mounted on de(.k with t h r u s t - m e a s u r i n g equipment. As
an -dternative to towing trials, some information can be
obtained from full-scale power trials, espe(.ially when the
l h r u s t T is measured. If we measure the l h r u s t - d e d u e t i o n
('oeflieient t on the model and assume there is no scale
effect on t in passing to the full-sized ship, then the resistance of the ship is

h',,
6

:/' (1

t)

o.oo~

dg
0,0o'o

o*oo*z 9

O.ooz{-

~,o ~

VL
h,,o7 ~ = -~-

I
B~I~ ~

~lO v

4~07

Fig. -I Frictional resistance coefficients for palnts--Taylor


Model Basin plank tests

If we also assume that the residuary resistance <'oeIL


ficient ('~ of the ship and model are the same, lhen Ill('
fri('tiomd resistance of 1he ship can be found from
//~, = R r -

R.

We m a y well question the v a l i d i t y of t hes~ a s s u m p tions. Until recently no full-scale measurements of I


have been made, but the Wageningen Vietory ship prog r a m suggests that there m a y be appreciable scale effect- for example, the value of t increased fi'om 0.22 for a
19.33-ft lnodel to 0.27 for a 74.1-ft model (10). I l e e e n t
JOURNAL OF SHIP RESEARCH

ir-'

5
LIIIIlillllll
ii!lll!iiiilinn
nnllnll iiiiiiilUlUnlUiU UlIIIIIIIIlUlIlU
illillllliiiiiiiiiiinlllllliim
uillulul nillllnuinu nmllllllnllllll
lllll
iiiiiiiiii
iiiii lUll IllUllUllllHIIIliiiiiUi!lllllllllllllllllll
i---I ~auin~
undiiiiimmm
mmmnUU nllllMIIIIIIIIIIIIII
~,--llau~
nu~ amiiillmm~diiiiiiiiiiiiiiiiiiiiilallllllllM
illliliilumnlniiiiiiiii
t flu nl n u UllUllllnllllll HlUlIIIIIIIIIII

i i ' '"'"'"'""""'
I

IIIIIIIIIIIIIIIIIIIIIIIIIIItlttllllllllfltllllllllllIIIIIIIIIIlill
I I I I I IIIit1111111111
I I I I I I I I I I I t I I I I IIIIIItlIIIIItlIIIIIIIIIIIIIttlIIIIIIIIIIlt/IIIIIIII~J~H_LLJ,]H~__
I IIIIIlllllllllllllllllllllllllllllllltltlllltllllllllllll~
t I I I I I I I I I I ltlt IIIII
I I I I I I I I IIIIIII
IIIIIIit111111t111111111tllttliit111111tt111111111]1
ILJ-L{4 LLLL
I IIIIII
Itllt11111111
IIIIIIIIIIIII I I I I I II IIIIIIit/111/

t I

,I

I I I

II

i i i I i i iiiiillllllltllllliltllillniliiilnliiillmlillimllll

IIIIIIIIIIIIII

I I I I

I I I Itlllllllllllllllllllllllllllllllllllltltllllllllllltllllllllllllllllllllllll

llillllllllllllll
lli/lillilllillil
lllllllnUlln

//illlltliill

I I I III//II_LII/IIIII

I I I I I I I I I I IIII!1111111111111111111!111.111111111t!111111
u uiiiiiiiuliuuliiiiimll
InuiiiilUHlUlIIIIlUlII
IInlIIIIIIIIIHHIHIlUUlIIIlUlUlUUlIIIIIIII

IIIIIIIIIIIIIt/!IIIIIIIIIIIIllJJlIIIIIIIIIIIIILL!ILI~U,]I][J]]]II]]]I]~Ii

IIIIIlllllllllllllillllllllttllltlllllllllll

t,]J I

~ICHtHttH4tHttttHtttHttlitiliH~dlHiill~

t I I I I I I t I II11111t11111111111111tlIIIIIIIIIiilNNr1NINITINI~TNITiIll
I I ! i i i I i I i I I I I I I I I I I I I I IIIIIIIIIIIIIINNNIIH111NNINI]IN11INNIIN11

-H44
-

- -

l!lttiltttl

--

t~tti

t ~ !'tfNthl

@
I I I

~ 1 1 1 1 1 1 1 1 t 1 1 ! 1 1 1 1 4 1 1 1

LlllJJJ_llNllllllll)lt)lllllll~t"i~'illflt#-.i~ili'il"l

IIII

l I IIIIIII

_ _

'

'1

I ill i',t t
iJt ]~1[
i t~
l'~

t'l~

#..

been calculated on the basis of the Schoenherr


modelfrictlnandline'shipaSSuming
the C#? is the same for

......

tlllll

lillll

IIIIII

I0 s

",,I

lllllil

Illllll
I

It1111I

IIIIItt

!I
, - ~~ + ~
I
]~t~t[[t]~l I~-t-~-H~ H-H~-H- ~i--t~P - - ~i ii.~i i lii ~ ..[ t ~ t t ! t [ J

rllmli~llll/llllllltlqll]
I

TTI1

'

fl

ilIIIIIIIIIIIIIItlIIIIIIIIIIGIIIIIiI~HII

/0

I I i I IIIitlltll

Fff.~

....

fllI'tt"tt"It't/T"I't
"

1i
"

,,,l!t
. .

il

,.t!1!1

t111

~o

full-scale power and towing trials on the USS Albacore


have indicated a similar variation of t, the rate of increase from model to ship relative to change in Reynolds
number being almost the same as in the Victory model
tests.
The assumption that (7~ is (,onstant in going from
model to ship at the same V//~/L value is also ope:/ t:)
question, since as indicated earlier in this discussion s,m~e
of the components of CR should show a Reynolds-munber
effect.
If the ship measurements do not include thrust, a comparison can still be made from the SHP but this introduces still further assumptions.
Subject to all the doubts and uncertainties mentioned,
the value of Cf for the ship can be obtained and compared with the smooth-ship value obtained from the
model using one or other of the extrapolation meth~ds
proposed. The difference is the so-called "roughness allowance coefficient," AC~,. It must be emphasized that
the values of AC~ so obtained will depend to a great extent on the actual method of extrapolation used, and -tre
to this extent subjective rather than objective.

"N
e~

Roughness on Actual Ship Hulls

.--

C~1 .~. ~..~ .~. ~

CI~

e~

i--

e-

w~._ ~

;~

~,

Results of such an analysis are shown in Fig. 5, using


the ATTC 1947 line and the Froude assumption. Particulars of the vessels are shown in Table 1. The average
values of A('~, range from +0.00015 to +0.00100 and for
each trial there is a considerable variation of AC~ over
the speed range. For United States merchant ships,
welded construction with some riveted shell seams,
painted with brands of commercial paint, the AC~ values
vary between +0.00015 and +0.0004, the increase in
frictional resistance varying from 10 to 28 per cent and in
total resistance from 5 to 12 per cent (11). Similar correlations between model predictions and ship trials, also
using the A T T C line, were given for a large number of
British-built merchant ships in 1954~ (12). For clean vessels with a shell 50 per cent riveted and 50 per cent
welded, the average value of A('~, for seven ships was
+0.00026. For all-welded shell, five ships gave a negative allowance averaging -0.00010, and three others
gave positive values averaging +0.00020. The over-all
average for this Ml-welded group was practically zero.
Fig. 6 shows that in these cases also there was considerable variation in AC~. for any individual ship as the
speed varied.
At th'st sight we might assume that these very small
and even negative values of ACF indicated that allwelded ship hulls with commercial-type paint had
reached ahnost the limit of smoothness. It is known
that to maintain hydraulic smoothness the roughness
must decrease in actual size as we go to higher Reynolds numbers, which would imply that the ship hulls
must be smoother than the models and planks on which
the extrapolation methods are based. Anyone with a
knowledge of actual hulls would certainly not subscribe
to such a conclusion. The truth is that AC~., apart from
JOURNAL OF SHIP RESEARCH

"0030

0030

~.0025
us

"0030

"0015

_o

.002S

ROUGHNESS ALLOWAN "E


= 0"0004

Y,

-4

"0020

U
Z

13 IANK~R,
I~, TANKER

_tZ_ .T~

C
e

0015

0 0 1 0

~-I

_ _a

TANKEq !

TANNER

TANKER
TANKER

O
d

(~
-OOIO ~ t t _ TANK[R
r~xt_gL~
_,~_~_
KR
e

g
_U

AR~.[R[
e
TANK~Rt
C
51
CARGOI
C
52 TANKERI
TANKERI
TANKERI

5B TANKERI
2 x l O I~
~S
3
(.~ All Riveted
(b) 25% We'Lded
( : ) 50% Welded

-OOOS

3"5

4'5

C,-values

necessary

I -S"

6
7
8
9
I0 ~
V L
REYNOLDS NUMBER, RR =--'~-"

(d) ?St Nelded


(e) 100~ Welded
Fig. 6

~@
II

ooos

_,~L

2x109

SHIP-MODEL CORRELATION (Iess than 20 day~ mat of dock)


to give agreement

between

ehp deduced

from power

measurements

and S c h o e n h e r r ehp

possible errors from scale effect and other assumpt.ions,


depends to a great degree on the method of extrapolation
used, as stated previously. Results of some recent
trials with Frmwh minesweepers illustrate this fact (12).
Taking tile results at n speed of 6 knots, and using first
the Froude assumption with the A T T C line, and then
the tlughes' proposal that all the resisl.a|we represented
by (~- scales with ('> ,>ing in turn the A T T ( ! line and
then the Hughes line, the values of AC~.. are +0.00015,
+0.00060 and +0.00064, respectively, the form factors
('v. ('u being 0.36 and 0.46 in the two littler cases. The
first two values of A('~,. represent increases of 7.:3 and 29.2
per cent above t.he smooth A T T C line. While there is
thus apparently only a relatively sm~ll g,'dn to he expected by further attent.ion to the smoothness of the hull
if we compare ship and model on the Froude assumption,
if the comparison is made on the basis that ('v,/C~,. is
constant for a given shape of hull, then the p,~ssibilities
of improvement appe:~r much higher.
We thus see that A('F, while it must include any necessary allowance for roughness, is essenlially a correlating
factor to halam'e the predicl.ed snlooth-ship resislan('e
against that of the actuafl ship. For a successful method
of prediction in practice, it is necessary for eauh estat)lishment t.o comp~'.re actual ship and model prediction
daia, however the-e latter m a y be calculated, and so to
build up sufficient k~,,~wtedge of the values of this correlaling factor A(:F for different types of ships and conditions of surh~ce that estimates of power for new designs
m a y be made with (.~mfidmlve.
W h a t of the Future?

The [ T T C a.1 it ~ meeting in Scandinavia requested the


Skin Friction Committee to prepare a proposal for conJULY, 1957

sideratio|l at the next Conference for a "frictional fornml'rt.ion based on 1Leynolds mmlber" which would be
"t/(lequate for prae.tieal ship design purposes."
The first requiremenl~ rules out the use of the Froude
coefficients. The or,her friction formulation at present
ac('vpted by the Conference for the publication of model
data is the A T T C 1917 line, originally due to Schoenherr. A perusal of the discussion at. the Conference in
Sca|~(li|m.via suggests that opinion there was divided between those who, in view of the large a m o u n t of research
presently in progress in this field, favored the continued
use of the A T T C line until more knowledge and experience had b',wn gained of the new proposed methods, and
those who believed that the time was still far dist~mt
whell any linality wouM be rea('hed and that. some new
single li,le could be found now which wouht be 'an improveinent over the A T T C line and by inferen('e still he
based upon the Froude assumption.
What Must Be Sought in Any Such Line?
It must be ~,onfessed t h a t at present we caroler w i t h

(,ert~linty predi('t the resistance of one model from the


results of another model on a different scale, where no
question of roughness obscures the issue. This, then,
would seem to be one req~dremetd we must loolc for in any
m'u' lieu ~ that it must lead to better correlation betweet~
Jamilics of geo.~im.~ than does ardt ea'i.~ting liT~e.
It has b0en stated in the past. by several experimenters
that the values of Cv for geosim models, analyzed by using the A T T C line, generally become smaller with increasing size. If this is so, and the extrapolation method
is to be based upon lhe Froude assumption
('i~

('~ @ ('w = c()llst


9

then a line somewhat steeper than the ATTC line would


he required over the range of small-model Ileynolds
n u m b e r ~ s a y for R ranging from 5 X 1@ to 5 X 106.
For example, it has been indieated that the Luc!! Ashbm
geosim models can be reasonably well correlated by using
the ATTC line with an addition of 8 per cent over the
entire range of R--this has the effect, of eourse, of increasing the slope of tile bast<' ATTC line. ()n the other
hand, Murray has shown that the ge<)sim pairs f o r a
mmll)er of models run at the Stevens Institute Tank and
at Taylor Model Basin correlate very well using the
original ATTC line 114).
Discussing the results of geosim tests ()n models of the
Victory ship, ranging in size from 2.728 to 72.75 ft
length, van l,ammeren has stated that the "Sehoenherr
mean line is in agreement with the results of the whole
model family," and that "William Froude's original assumption that the speeifi(, residuary resistance in the
nonwavemaking region is constant, seems, therefore, t,)
be justified for the Victory model family" (15). An
analysis by Gertler of a large number ()f resistance experiments on a family of bodies of revolution, covering a wide
range of length-diameter ratios, and run deeply submerged so as to t)e free of wavemaking, has shown that
tile resistance curves are pvxallel to the ATTC line tip to
Reynolds numbers of the oMer of 2.6 X 107 116).
It will be seen, therefore, that there is a considerable
amount of eviden(.e to support the ATTC line as a pract,ical one f o r use in the correlation of geosim nn)dels if w('
are looking for a single-line extrapolator.

Extrapolator for Each Model Shape


If we waive this latter re(luirement , then we have to
consider the proposals which have been made on the basis
that each model shape requires a unique extrapolator.
The principal advocates of this approach are Telfer, Lap.
Troost, and Hughes.
Telfer has worked for many years on this problem,
and believes that the results of any geosim series can be
('orrelated hy a straight line when Cr is ph)tted against
R - v ~ Lap-Troost, in their formulation, state that the
correlation can be obtained by variations in the abscissa
when Cr ix ph)tted against log R, a term in the fornmla
being of the form of log A, where :1 is a coetfi('ient--depending upon the actual hull form. Hughes advocates
the use of form factors (Y~./('~,, which will be constant for
any given form.
In theory, it should be possible to distinguish between
these different proposals by applying them in turn t() the
existing geosim data. in. practice, this is extremely diffi('ult. At the, h)w V %/L speeds at which wavemaking
is M)sent, it is difficult to measure the resistance ac('urarely in the first place, and therefore to determine the
correct slope of the model curves in this region at what
Hughes ealls |,he "run-in" point. At, the h)wer end of the
curve, also, the results are often suspect for lack of ade(luate turbulence stimulation, while at the higher end
interference effects between the model and the sides
and t)ottom of the tank enter into the problen~. Thus
10

in general the results ovel* only a very limited speed


range for any one model are avaihd)le for such correlation
analysis.
Despite these practical difti(.ullies, the Internatiomd
Committee has made a determined effort to glean any
iifformation possible from existing geosim data. Most
of the material available has been tabulated at Taylor
Model Basin, and a co-operative effort made t') analyze it
by the various proposed methods. At the request ~ff
the committee, Hughes, Lap, aml Telfer undertook to (h)
their own analysis by their respective methods, and \rill
doubtless present any results they may wish to the Conference ill Madrid. At T-~ylor Model Basin the analysis
was vonfined to the use of the A T T ( ' line, and the results
were presented in detail by Hinterthan to the ATTC in
Washington in 1956 (17).
The analysis was carried out in the same way as that
done by Murray 114)-- the ('xe values for the largest model
of the series were assumed to be the most nearly eorre('t,
and then subtracted, at the appr()p riat e 1" %/L values, from
the Cr values for the snmller models, so obtaining values
of ('~, for these smaller models. The analysis covered
some 2(; geosim series (in<'luding pairs ,)f models). The
results showing all the derived sI)ots were given in reference 117) and there is evidence that many of the
models were affected t)y htminar flow. In order to
eliminate some of this effe('t and also of wall effect at high
speeds, "~ second plot was nmde to include only the speed
range from 10 per ('ent below to 5 per ('eat above the
service sl)eed, the latter being defined by Troost's f,,rmula 118)

Vs
~//LBI,

1.85

--

1 . 6 (~r. . . . . . . . . .

[31

--

The restllts of this anMysis indiea|ed that greater care


is necessary ill the testing of the models to eliminate
various inaccuracies ill measurement, turbulence problems and wall effects, hi general, the over-all picture
suggests the need for a ('on'elation line somewhat steeper
th-m the ATTC line over tile lower range of IKeynohls
mmd)ers, say below los ~, R = 6.6.
()ver the ship range of l{eynol(ts numbers, n() basic experiment:d data are available to determine the sm()oth
turbulent curve. As has been pointed out in the foregoing, full-scMe trial data only lead t() values of an all()watwe t'at't,)r AC~., which must include not only the tilldoubted effe('ts of roughness and fouling on the ship, but
also all the errors in tile sealing laws applied to the model
data in order to obtain the smooth-ship prediction. If
we use the ATTC line for extrapolation to the ship and
the Fr,~ude assumption of ('onst,an(.y of ('e, it h~s been
show~ th:~t trials of modern, all-welded ships lead to
small, zero, ()r even negative values of A(~,. This at
first would suggest that either the ship hulls are considerM)ly sm()other, in the absolute sense, than the surfaces
()f the models, or that the ATTC line is too high in this
renge ()f ll%qml(ts mmfl)er. Th(' fornler is untenahh ~, and
the latter is only necessarily true if the Froude assumpJOURNAL OF SH|P RESEARCH

tion holds. If, in fact, the value of (,'~ is less at ship values of R than at, the model values for a given hull forln,
as suggested by Hughes and Lap, then the same trial
data would lead to considerably increased values of AC>
as just shown for the ease of the French minesweepers.
At. this time we have no means of separating these vari()us effects, and whatever method of extrapolation is used
lo predict the smooth ship resistance from that of the
smooth model, the final step to the actual ship can only be
taken t)y the use of an empirical allowance of the t.ype
A(,*> If this factor is (;ailed, for convenience, a "roughhess allowance," it is well to rememt)er that the name
covers a multitude, of sins in the form of our ignorance of
1he fundamentals of the problem.
At present, the results of all model resistance tests
published in reports and papers are either based upon
lhe Froude or ATT(! formulations, and the labor entailed in converting one to the other is very great. Oi'tel~
it is never done, and large amounts of information are
lost to the busy naval archite(,t dealing with resistan('e
problems. The f('eling~ so far as N o r t h America is concerned, was reflected by the action of the Americ:m Towing T a n k Conference at its meeting in ,qeptember, 1956.
After much discussion, ranging over all aspects of the
subject, t.he Conference unanimously adopted n resolution put forward by its Skin Friction Committee, under
the (?hairmanship of Dr. L. Landweber, to the effect that
"the committee has reviewed the present state of knowledge of the sut)iect and in view of the great amount of
work going on in this field feels that the tilne is inopporIune to Inake any departure from the decision to use the
1947 A T T C line."

Conclusions Reached at TMB

The entire prohlem has been reviewed carefully at the


Taylor Model Basin over the past three years and the
results of m a n y geosim tests studied. The general conch>
sions from this survey were well set out b y Granville in a
recent paper (19), where he stated:
1 The basic Froude hypothesis concerning the separation of viscous and wavemaking resistances is justified.
2 The data are not accurate enough to establish the
best procedure of extrapolating form or viscous resist"HlCO.

3 It is imperative that a full-so,de ship with a hydrodynamically smooth hull be tested.


4 Additional res('ar('h is needed into basin-wall effects on the large models and the stimulation of turbulent
flow on the small models, the positi(m of transition on th(,
model to be experimentally determined so -~s to eliminate
troy doubt on this score.
The use of a flat-plate line us a reference line for form or
viscous resistance still seems a fruitful method of inquiry.
The concept of the forin or viscous resistance coefficient h'tving a fixed part and a part proportional to the
flat-I)latc resistan<,e coefficient is inclusive enough to enJULY, 1957

compass any probable variations. Finally, more accuracy is needed in the experimental procedures before any
f n a l decision as to the correct method of extrapolating
model results to ship predictions can be reached.
The Taylor Model Basin has t)een using the A T T C line
since it.s adoption in 1947. The surface ship models
tested for resistance and propulsion are normally between 20 and 30 ft i~t length. The ship results are predicted using the A T T C line, and fi'om the fuli-scale trials
values of A(.'~., have been obtained for a great variety of
types of ships having different types of structural roughness an(1 different paint surfaces. Bearing in mind that
AC~, is in fact a "correlating factor" and not a roughness
allowance, we are well satisfied with the present system
as an engineering method, though fully aware of its defieien('ies in the fundamental sense. We have accunmlated a vast, amount of data over the t)asl, 10 years, an(t
have no desire to reanalyze these and <)blain new "('orrelating fa<.t<~rs" f<)r future use unless assured of no further change for a number of years. In view of the Mmwn
limitations of the Froude t~ssumption and the use of a
single-line extrapolator and all the new work in progress
on form effe('t, any su<,h assurance seems impossible today. Also, any method involving the use of form factors
necessarily must go through a trial period before being
adopted univers'dly. We would re(tuire considerable experien('e with any system using the met hnd of "rim-in"
values of C~,. to determine the form factor before being
willing to change to such a system.
The Model Basin is aware, however, ()f the difficulties
of the extrapolation of resullas fi'om small models, and
would be prepared to go along with some steepening of
the extrapolation line in this range of t{eynolds numbers
should the International Conference decide to recommend
it.
Conclusion

In view of the m a n y areas in which our knowledge is


still so imperfect, and of the great a m o u n t of research
presently being carried out in the fields of laminar and
turbulent b o u n d a r y layers, roughness codification, and
geosim analyses, both b y theoretical and experimental
methods, no finality is in sight, at this time. However, it
would be a major st,ep forw'~rd if all establishments could
agree on a single procedure, so that at least new published
data would be at once comparable. At the same time,
if the proposed proce<ture is not one of the two already
accepte<l, then it will need to have s()me strong claim to
temporal permanence for a number of years ahead before
the different experiment tanks are likely to view with enthusiasm the change to a new system and the consequent
difficulties of assimilating the results into their already
('onsi(lerable archives.
Bibliography

1 "Friction and F o r m Resistance in Turbulent Flow,


and a Proposed Formulation for Use in Model and Ship
Correlation," by (I. Hughes, [NA. 1954.
11

2 "The Frictional Resistmu:e of Flat lqales in Zero


Pressure Gradient," by I,. I~andweher, Trans. SNAME,
vol. 61, 1953, pp. 5-:21.
3 "The Problem of t.he Turbulent Boundary l,ayer,"
by D. Coles, ZAMP, 1954.
4 "Turbulenl Friet.ion on a Flal. Plate," by A. A.
Townsend, 7th International Conference .n Ship Hydrodynamics, Oslo, Norway, 1954.
5 "Ship th~sistance Simihrrit.y," by E. V. Teller,
INA, 1927.
(i "Frielional Drag of Ship Forms," by A..1. W. Lap
and L. Troost, SXAME, 1952, Northern (:alifornia Se~'lion, Soeie W Bullet.in, June, 195:1.
7 "Einfluss der form auf den reibungswiders/m~d,"
by (I. Wigley, Internationale Tagtmg der l~eit.er der
Sehlept)versu('hs:mslalten, Berlin, Germany. 1937.
8 "Effe(,t; ()f Transverse (flu'vagure - , Fri('t.ional
]{esist;anee," by 1~. L~mdweber, Taylor Model Basin l{eport 689, 1949.
9 "Skin F r i d i o . l'esislalwe a n d t,he Effeds of Surface l{oughness," by F. 1I. Todd, Trans. SX.\ME, voI.
59, 1951, pp. :-115, 368 37k
10 "Scale Eft'eel, Experimenl.s on Viebory Ships and
Mode, Is," by W. P. A. wm I,ammeren, J. D. van M'men,
and A. a. W. Lap, INA, 1955.
]1 "A Review of ()ur Present Knowledge on Skin

12

Friction l{esistan('e," t)3" F. 1t. Todd, Sehiffsteehnik,


November, 1955.
12 "Ship Trial Performamu' and the Model Prediction," by J. F. Allan and [t. J. S. Ctmhtml, [NA, 1954.
13 "Etude a t a m e r de la resistance "~ la mar,he et de
la propulsion-rapprochement a.vee le nmdele," by 1{. l{etall and ,~. Bindel, A.T.M.A., 1955.
14 "Frictional [{esistance CoefIi~dents at Low Iieynolds Numbers ()blained by (].mp-~ris(m of I~a'ge "rod
Small Models," by A. B. X[urray, 10th Meelil,g of
A.T.T.C., 195:k
15 "S("~lo Effect Experimm,ts oil \idory Ships trod
Models," by W. P. A. v ' m IAt_IIIIIK'F(HI, ,J. 1). Villi 5[,~tllel],
and A..l.W. Lap, I X 5 , 1955, p~rt [, appendix Ii, p. 2:{2.
l(i ";\pplieation of the L~q>Troost Exlrapoladon
Method to SubmerKed Bodies of lh<olution," I)3" hi.
(lertler. 10th 5.Ieeti~g of A.T.T.C., 1!15:{.
17 'ql, eport on (;eosim Amdy,~is According to Schoenherr IAne," by W. B. Ilinlerthan, 1 llh M~'eting of A.T.T.C., 1956, and T M B Report No. t064.
18 "A Simplified Melhod for Prelimimlry Powering
,ff Single-Screw Mereh-mt Ships." 1)3- l~. Troost, S N A M E
New England Seal.ion. 1955.
19 "The \Tis,:ous ll.esistam'e of Sm'h~ce Vessels mid
lhe Skin Friction of Flat Phd, es," by P. S. (Ir'mville,
Trans. SNAME, \ol. t;4, 1.956, pp. 209:227.

JOURNAL OF SHIP RESEARCH

Vous aimerez peut-être aussi