Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
c; Local frictional resistance coefficient for turbulent which the water produces on the hull surface or skin of the
flow. ship. This generally varies in a regular manner with speed
d Diameter of solid of revolution or thickness of two- and, apart from the viscosity and density of the water, it
dimensional body. depends primarily on the area, length and roughness of
f Froude’s coefficient of skin friction in RF =fSVn. the wetted surface of the hull. It is that part of the ship’s
g Acceleration due to gravity. resistance which to a great extent is uncontrollable in design,
k Diameter of sand grain. except in so far as the wetted surface is affected by choice
L Length of ship or body, in particular, length of full- of proportions.
scale ship. The wave-making resistance manifests itself in the form
Lc Distance of transition point from leading edge. of normal pressures on the hull due to the surface wave
1 Length of model as distinct from ship. systems set up by the action of the hull on the water. This
m Power of the Reynolds number at which the part of the resistance is the summation of the components
resistance coefficient varies C oc R,m. in the direction of motion of these normal pressures. Wave-
n Power of the speed at which the resistance varies, that making resistance can vary in an undulatory manner with
is, R oc V”. speed due to interference effects between the bow and stem
PS Shaft horsepower. wave systems. Apart from the density of the water, it is
P Index of the power law for velocity distribution in primarily affected by geometrical features such as propor-
the boundary layer v/V = (y/S)l/p. tions, fineness and shape of hull and is that part of the ship’s
4 Ratio of skin friction to total resistance, RF/RT. resistance which can be controlled in design.
R Total resistance. The third component, that is, resistance due to eddy-
RF Skin friction resistance. making, has usually been associated in the past with energy
RW Wave-making resistance. losses due to the formation of eddies behind bluff endings
Rn Reynolds number on total length, Rn = VL/v. in way of the stern of the ship; more especially those as-
R, Reynolds number on distance from leading edge, sociated with the rudder and stern post and the propeller
R, = Vxiv. shaft supports. As explained later, however, separation of
Rc Critical Reynolds number. flow leading to the formation of vortices and eddies can also
r Form factor. take place on streamlined bodies under certain adverse
rW Wave-making resistance of model as distinct from ship. pressure gradients associated with the contracting flow in
S Wetted surface area. the after-body. This resistance also manifests itself in the
V Speed in general. Free stream velocity, in particular, form of normal pressures on the hull but is generally of
speed of full-scale ship. viscous origin. By good design this form of energy loss can
V Speed of model, as distinct from ship, speed within usually be reduced to negligible proportions.
boundary layer. The fourth component referred to is the resistance due
W Weight density, pg. to the relative wind acting on the above-water part of the
CL Coefficient of dynamic viscosity, shear strcsslunit ship. Owing to the very low density of the air compared
velocity gradient. with water, this is generally quite small compared with the
X Distance from leading edge. water resistance and is usually not very significant except in
Y Distance from surface. quite strong winds.
U Mean hull roughness. In rough water there is also additional resistance due to
6 Thickness of boundary layer. ocean waves and ship motions, but this aspect is beyond the
h Linear scale ratio, LII. scope of the present lecture.
V Kinematic viscosity. For a well-designed hull form, therefore, it can be said
P Mass density. that there are two principal components of resistance,
Pm Density of water for model. namely, skin friction and wave-making. By dimensional
PS Density of water for ship. analysis these can be expressed in terms of non-dimensional
7s Surface shearing stress. parameters as follows :
4 Complex function depending on the proportions and
geometry of the ship.
broad breakdown of ship resistance since the subject was T o satisfy the Reynolds law, however, the product of
first rationalized on these lines by William Froude about a speed and length must be the same for model and ship,
century ago, it is now realized that the position is not quite that is, the speed must vary inversely as the length. The
so straightforward and clear cut. These aspects and the speed of the model would then have to be greater than that
assumptions involved in the practical evaluation of the of the ship, namely, v = AV. The requirements of the two
various components w ill now be discussed in the light of physical laws are therefore quite different and, for a given
recent developments. fluid medium, they cannot be simultaneously satisfied in
a small-scale model test. This impasse, although not fully
THE B A S I S O F THE MODEL TEST understood at the time, was ingeniouslyovercome by William
Although considerable progress has been made in recent Froude, the great pioneer of ship model testing as we know
years in the development of the mathematical theory of ship it today (I)*, by assuming that skin friction and wave-
resistance (incidentally this generally refers to wave-making making resistance were for all practical purposes indepen-
resistance only) the position is such that ship designers dent and separable. This amounted to expressing equation
must still rely in one form or another on the results of tests (1) in the following form:
on small-scale models in experiment tanks for the powering
of ships. In other words, although theory has shed much
light on the general nature of the # function referred to
earlier, the solutions are only tractable under quite radical where #, and +2 are independent functions, that is,
assumptions and for actual ship forms they can be reliably
determined only from experiment. In this connection, tank
tests are used in respect of the following:
Froude made the model dynamically similar to the ship
(1) the comparison of the relative merits of different
so far as wave-making resistance was concerned, by satisfy-
ship forms to suit specified requirements.
(2) the prediction of the full-scale resistance for the ing the non-dimensional parameter V d g T and regarded the
chosen form. skin friction as something apart for which separate estimates
would have to be made for both model and ship. He based
Whereas the comparisons referred to in (1) are generally these estimates on skin friction data, obtained from compre-
made on the basis of model tests alone, the predictions hensive tests he carried out on a series of planks of different
referred to in (2) are, at the present stage of development, lengths which were towed in his first experiment tank at
made in conjunction with a factor of experience or ‘corre- Torquay (2) (3)-
lation factor’ referred to later. T o put this into effect Froude then made a second major
Although it is beyond the scope of this paper to deal at assumption, namely, that the skin friction of both the
any length with the technicalities of ship model testing, it model and the ship would be equal to that of a plane surface
is relevant to refer briefly to the background to this, at least of the same length and area as the wetted surface of the
so far as the prediction of ship resistance is concerned. model and ship respectively. In the light of present-day
Indeed, in considering the skin friction of ships it will be knowledge it can readily be shown that this assumption was
seen that it is appropriate to approach the subject from the not quite correct, but it was a very reasonable one at the
model end, so to speak. time and is, in fact, still being effectively used.
I n considering the fluid flow around a ship and a small- Froude’s procedure was then to measure the resistance
scale model, complete dynamic similarity could be obtained of a small-scale model in his experiment tank at a series of
only by simultaneous satisfaction of the Froude and Rey- speeds and subtract from this the estimated skin friction
nolds laws for wave-making and skin friction resistance from his plank data. The remainder, or ‘residuary’resistance
respectively, that is, the dimensionless parameters ~ / d s as he called it, was then taken to be largely composed of
and VL/v should have the same value for model and ship. wave-making resistance corresponding to the function 42
For a given hull surface condition the dimensionless resist- in equation (2) which at ‘corresponding speeds’ would be
ance coefficient RIpV2L2 would then have the same value expected to have the same non-dimensional value for model
and the full-scale resistance could be readily determined and ship. This would mean that
from the measured model resistance. psV2L2
For example, if the linear dimensions of the ship are A R -r -
- wpmv212
times those of the model, then to satisfy the Froude law the
model would have to travel slower than the ship at a speed where Rw and rw are the wave-making resistance for the
proportional to the square root of the model and ship ship and model respectively and p s and p m are the densities
lengths, that is, of the water in each case. As shown earlier, however,
V = A4v and A = L/l so that
R , = r W Ps
rA3 . . . . (3)
where the capital and lower case symbols refer to the ship Prn
and model respectively. * A niimerical list of references is given in Appendix III.
Proc Instn Mech Engrs Vol I76 1962
It will be seen, therefore, that for the same density of water the fluid apart from the viscous shearing of consecutive
for model and ship the wave-making or residuary resistance fluid laminae. In proceeding downstream from the leading
varies as the cube of the linear dimensions or as the volume edge of the plane the laminar boundary layer gradually
of displacement. Taking the densities into account this increases in thickness and, eventually, at a critical value of
simply means that the residuary resistance varies as the Reynolds number VL,/v (where L, is the distance from the
weight of water displaced. Or, as Froude put his law of leading edge), the laminar flow breaks down and becomes
comparison : the residuary resistance per ton of displace- ‘turbulent’. The fluid particles then have an oscillatory
ment is the same for model and ship when running at motion about a mean flow path instead of the strictly
‘corresponding speeds’, that is, speeds proportional to the rectilinear motion appropriate to laminar flow. As the
square root of their lengths. boundary layer is much thicker, the momentum generated
It should be mentioned here that at the time Froude’s and consequently the resistance to motion is considerably
law was enunciated in 1870 he was not aware of the precise greater. A pictorial representation of this development
form of the viscous parameter VL/v, but from his argu- along a plane surface is shown in Fig. 1. A point of particular
ments and reasoning there is no doubt that he was fully interest is that, even in turbulent flow there is still a very
aware of its implications in general terms at least. Reynolds’ thin laminar sublayer through which the final transfer of
treatise dealing essentially with fluid flow in pipes did not momentum is made.
appear until some time later, in 1883 (4). In Fig. 2a is shown a comprehensive resistance plotting
for these different flow regimes on plane rectangular surfaces.
The ordinate is the non-dimensional resistance coe5cient
THE NATURE O F SKIN FRICTION in the form CF = RF/+pV2Splotted on a base of Reynolds
It is convenient in the first place to consider the skin number R, = VLlv, where RF is the skin friction resistance.
friction of a flat surface moving in its own plane, just as Compared with the resistance coefficient in equation (I)
Froude did. it will be seen that the size characteristic L2 has been
As pointed out by Reynolds (4) in connection with flow replaced by S, the total surface area of the plane in contact
in pipes, there are at least two flow regimes in this work: with the fluid, and the numerical coefficient of has been +
that known as ‘laminar flow’ where the viscous forces pre- introduced into the denominator. The coefficient now has
dominate over the inertia forces and that known as ‘turbu- the physical significance of the ratio of the mean shear
lent flow’ where the converse is the case. The former is stress on the plane (RF/S) to the dynamic or ‘stagnation
generally associated with Reynolds numbers less than a pressure’ of the fluid (+p . V2).Consequently, the numerical
certain critical value and the latter with higher values. value of the coefficient is of the same order of magnitude
For fluid flow along smooth planes there are generally as the ratio of the resistance of the surface to motion in its
three cases to be considered: own plane to that when the plane is at right angles to the
direction of motion or ‘head on’ to the flow. This explains
(1) purely laminar flow;
(2) transition or ‘mixed flow’ where the extent of the why the numerical values of the coefficient C , are so small.
The upper bold line AB is the resistance coefficient C,,
laminar region is of the same order as that of the turbulent
region; for turbulent flow, due to Prandtl and Schlichting ( 5 ) and
(3) purely turbulent flow or where the extent of the the lower bold line CD is the resistance coefficient C , for
laminar region is relatively small and its effect negligible. laminar flow due to Blasius (6). This shows clearly how,
for a given value of the speed parameter VL/vlaminar flow
Laminar flow is less resistant than turbulent flow, the width is much less resistant than turbulent. Also of interest is the
of the frictional belt* or boundary layer normal to the sur- fact that both the resistance coefficients for turbulent and
face being much thinner and there is little disturbance in laminar flow, reduce with increasing R,. This means that if
That part of the jluid in the immediate tiekhbourhood of the plane we regard R, as being increased due to length alone, the
whose motion is affected by the shear stress at the surface. specific resistance or resistance per square foot becomes less.
I TRANSITION i
Y-SPEED INI
FREE STREAM
c
I
LAMINAR SUBLAYER
I-DISTANCE FROM
LEADING EDGE
- LEADING EDGE- J
Fig. 1. Schentatic development of boundary layer on a frat plate: Vertical scale 6 corresponds approximately to 20
times horizontal scale x
Proc Instn Mech Engrs C’ol 176 I961
u)
(1
%
a
-yu
\
k
0
i
I-
* z
w
0
L
L
w
ou
W
0
z
f
-
v)
v)
w
a
-I
a
z
0
c
0
a
Lr
2
a
s
090;
(1
L Ooot
0
Lk
\
I.
II
-6QOO!
c
z
w
Y
; 0004
8
w
0
z
8 0.003
E
an
W
a
-I
9002
E
0
K
LL
0.001
<
0
0
4
0
1
LOCAL REYNOLDS NUMBER. R x = V X / v
b Local.
Fig. 2. Curves of frictional resistance coeficcient
Proc Insrn Mech Engrs Vol I76 I962
In other words, so far as skin friction is concerned, increase Also shown in Fig. 2a, above the turbulent friction line
in length is generally an advantage and this is explained by AB are a series of curves representing the resistance of a
the fact that any extension of a plane surface at the down- particular form of rough surface. These correspond to the
stream end benefits from the forward-moving fluid, the surface of the plane being completely covered with tightly
momentum of which has been generated by the surface packed sand grains of various sizes. The significant para-
ahead of it. meter for these is the relative roughness k/L, where K is
Referring now to the series of curves marked ‘normal’ the grain size and L is the length of surface. These curves
transition curves such as gh, these represent the usual form were developed by Schlichting(8) from tests carried out
of transition when no artificial means are taken to stimulate by Nikuradse with roughened pipes and will be referred to
turbulence. Each curve corresponds to transition at a con- later. For a given value of the roughness parameter K/L it
stant critical value of Reynolds number R, = VL,/v (for will be noted that:
example, R, = 106 for the curve gh, that is, the position at
which the curve meets the line for laminar flow). It then (1) U p to a certain value of the Reynolds number the
follows that the extent of the laminar strip at the fore end resistance is identical with that of the smooth plane in
decreases with increasing speed (the product VL, being turbulent flow.
constant) and the transition curves gradually approach the (2) As R , increases the resistance coefficient rises
fully turbulent line C,. It can be shown (7) that such tran- above that for the smooth surface and eventually attains
sition curves take the form : a constant value (resistance varying as speed squared).
K This behaviour has been explained by the fact that up to a
c, = (2,-- Rn
. . . (4) f
certain speed the roughness elements are completely en-
veloped by the thin laminar sublayer and, as a consequence,
where K is a constant depending on the critical Reynolds are not resistance-sensitive. At higher speeds this sublayer
number R,. Although transition curves of this form are becomes progressively thinner, the roughness elements then
deduced from theoretical considerations based on certain project through and cause additional turbulence. It is
assumptions, there is considerable evidence to coniirm their stressed that this behaviour is not necessarily general, but
general nature (7). Experience also shows that the point at applies particularly to sand roughness. It was considered
which transition takes place can vary according to circum- appropriate, however, to refer to it briefly at this stage in
stances as, for example, the presence of initial turbulence view of its importance. Of particular import is the fact that
in the fluid or the condition of the surface. I n other words, for minimum resistance a surface does not have to be per-
there is not a unique transition curve but, even so, when no fectly smooth and a certain amount of roughness can be
special means are taken to stimulate turbulence an average borne without detriment in that respect. Ths is known as
figure for the critical Reynolds number R, appears to be the ‘admissible’ roughness and is plotted in Fig. 3.
about 4.5x 105. This makes the constant K i n the transition The above considerations refer to the skin friction of a
curve 1500 as shown in Fig. 2a. Curves corresponding to plane as a whole and the coefficients CF,C, and C , are
transition at R, = 2x 10’ and 106 are also indicated. usually referred to as the ‘total’ or the ‘mean’ resistance
Between the two lines AB and CD (i.e. complete turbu- coefficients. An important concept in this work, however,
lence and laminarity respectively) are a series of inter-
mediate lines of somewhat similar character which also give
resistance coefficients for mixed flow conditions. These
correspond to lixed degrees of laminarity between 0 and
100 per cent; for example, the line ef marked 30 per cent
laminarity corresponds to the first 30 per cent of the surface
area from the fore end being subject to laminar flow and
the remaining 70 per cent to turbulent flow and so on over
the range. These correspond to the stimulation of turbulence
at fixed positions along the length by special devices.
It is also to be pointed out that there is no reason to
suspect that the various relations discussed above do not
apply to all fluids including both air and water provided
due cognizance is taken of the appropriate values of p and v.
Indeed, all the evidence there is tends to confirm this
generality.
So far as water is concerned the effect of temperature on
p is small, but v is affected quite significantly. For turbulent
flow on a smooth surface the net effect is that there is a .-
REYNOLDS NUMBER. R n n V L / v
reduction in resistance of about 24 per cent for a rise in 4
temperature of 10°F. Some figures for this are given in Fig. 3. Schlichting’s admissible roughness for vurioiis
Appendix 11. lengths of plate
Z+oc Insrn Mech Engrs I‘oi 176 I962
is that of the ‘local‘ skin friction or the actual shearing For fully developed sand roughness (R E Vz) the resist-
stress T at a point in the surface, say, distance x from the ance coefficients are given by:
leading edge. This is defined by
r
CF’ = -
+p v2
where CF‘ is the corresponding dimensionless resistance 2.87- 1.58 log,, - -
X
]
k -24
. . (14)
coefficient known as the local specific resistance. It can be
Fig. 4a shows typical velocity distributions in the
shown (7) that provided the mean specific resistance CF is
a function only of Reynolds number there is a definite boundary layer for both laminar and turbulent flow where,
relationship between the mean and local coefficients, for convenience, the plane is at rest and the body of fluid
is flowing past it at speed V. These are non-dimensional
namely, presentations in which the fluid speed z, at distance y from
CF’=R,-+CFdCF . .
dRn
(6) - the wall is expressed as a fraction of the free stream speed V,
and y itself is expressed as a ratio of the boundary layer
If n is the power of the speed at which the resistance varies thickness 6.
at given R, then it can be shown that The distribution for laminar flow in this form is unique
for all values of R, and therefore any position on the plane,
(7) although, of course, the actual thickness will vary. On the
other hand, the distribution for turbulent flow varies with
then from equations (6) and (7) R, as shown. Some distributions for rough surfaces are
CF’ = cF(n-1) . . * (8) shown in Fig. 4b.
This gives a very convenient relation between the ‘local’ For both types of flow v = 0 at y = 0, that is, there is
and ‘mean’ specific resistance at a given point. For example, no ‘slip’ at the fluid-solid interface. This is generally
in purely laminar flow the resistance varies uniformly as accepted to be in accordance with the experimental evidence
speed raised to the power of 1+ which means that the local and is referred to later.
coefficient of resistance C,’ is exactly half that of the mean Fig. 4c shows typical laminar and turbulent velocity dis-
coefficient C,. tributions in a pipe and their resemblance in character to
For turbulent flow the speed exponent n is not constant, those for the flat plate will be noted. Indeed, much of our
but increases slowly with R, and is always less than 2. At knowledge of the nature of boundary layers has arisen as a
R, = 107, n = 1.84, which means that at this point result of the study of pipe flow.
C,‘ = 0.84 C,. The nature of boundary layers is discussed in more detail
The broken line A’B’ shown in Fig. 2a is the local in Appendix I where methods of calculating their thickness
resistance coefficient for turbulent flow. It will be noted are also given.
that the ‘normal’ transition curves all intersect the line
A’B’ at their maxima which is a characteristic of these FROUDE’S SKIN FRICTION EXPERIMENTS
curves (7). In view of their importance in the development of the
In Fig. 26 are shown the curves of local resistance co-
subject of ship resistance, a very brief account will now be
efficients CF‘ corresponding to the mean values in Fig. 2a. given of Froude’s classic experiments on the surface friction
The abscissa is now the local Reynolds number R, = VX/V, of planes. These were carried out in the year 1872 in the
where x is any distance downstream from the leading edge. experiment tank he had built near Torquay shortly after he
The transitions from laminar to turbulent flow corre-
had laid down his law of comparison for model tests. The
sponding to the ‘normal’ curves shown in Fig. 2a are results were published in two reports to the British Associa-
indicated by the vertical lines such as ab. There is evidence tion (2) and (3) together with a tentative suggestion as to
to suggest that transition is, in fact, more gradual (7) but how to use them for estimating the skin friction of planes
this does not affect the general shape of the mixed flow of length greater than the longest plane tested.
curves once transition has taken place. The planes were wooden boards 19 in. wide and varied
The principal relations illustrated in Figs. 2a and b are
in length from 1 to 50 ft. They were towed in a vertical
as follows: plane in fresh water, completely submerged, at speeds up
0.455 to 1000 ftlmin. A number of different surface finishes were
‘ - [log,, Rn]2.SS . . . . (9)
C -
tried but the principal one was the varnished surface, which
was regarded as the basis of comparison.
C,’ = c, 1-
[ l,;1g: The concept of Reynolds number for the analysis of
surface friction was not known at the time and Froude
1.328 analysed his results empirically on the assumption that skin
C -- . . . . .
I - R,* friction would vary jointly as the wetted surface area and
0.664 the speed raised to some power, that is,
C1’ = - . . . . .
R,+ RF = f S P , where f and n were constants.
Proc Znsrn hfech Engrs Vol I76 I962
He found this held quite well for a given plane but thatf
and n did, in fact, vary from one plane to another and, in
particular, tended to decrease with increasing length. By
noting the nature of the decrement off, values off were
extrapolated for lengths beyond that of the longest plane
tested. The value of n varied very slowly with length and
was assumed constant in the extrapolated region and
approximately equal to that for the 50-fi plane. This was
eventually fixed at 1-825throughout and the f values were
adjusted accordingly. A detailed account of how this extra-
polation was made for ship lengths up to 600 ft and its
subsequent extension to lengths of 1200 ft was given in a
paper by Payne in 1936 (9).
In reference (7) Froude’s original experiment results were
re-analysed on a Reynolds number base, the essence of
which is shown in Fig. 5. It is pertinent to remark here that
whereas Froude was eminently aware that wave-making
v/v
and skin friction followed quite different laws he was not
in a position to know at the time that skin friction itself
a Smooth plate.
was subject to different flow regimes. In this connection it
is quite clear from Fig. 5 that all Froude’s planes of 1 6 4
length and less were suffering from laminar flow to varying
extents, many of them following well-defined transition
curves. Indeed, it appears that only the 28- and 50-ft planes
were reasonably free from this effect.
Also shown in Fig. 5 are values of specific resistance
corresponding to Froude’s empirical analysis and extra-
polation referred to earlier. It is clear that these do not
conform to Reynolds law in that CF is a function of both
Reynolds number and surface length. Values are shown for
various constant lengths varying from 5 to 1000 ft, the
specific resistance increasing with length at a given value
of R,.
It is clear, however, that when the results are analysed on
the more rational Reynolds number and due account is
taken of the laminar flow on the smaller planes, Froude’s
results are, in fact, compatible with a unique turbulent
friction line* such as that shown in Fig. 5. This, as described
“lV in (7), is also supported by other experimental evidence-
b Plates with various degrees of roughness. in particular, Kempf’s measurements on painted steel
surfaces ( I 0 ) and ( I I).
Towards the right-hand side of Fig. 5 it will be noted
that Froude’s extrapolated values of resistance rise above
the turbulent friction line, and the more so as length
I increases. This is the region of ship Reynolds numbers and
MEAN SPEED
-
OF FL0W.V
-
-- , i
i CENTRE-LINE OF PIPE 1 1
it is often claimed that Froude included some allowance for
hull roughness for these longer lengths. In the author’s
opinion, however, a very significant factor in this gradation
with length was the undoubted fact that Froude’s shorter
planes were suffering from laminar flow and offered rela-
tively less resistance on that account. This must have
affected to some extent the length effect arising from the
empirical analysis which doubtless made itself felt in the
extrapolated region.
The capricious effects of laminar flow on skin friction are,
of course, now well known but, as will be seen later, the
c Smooth pipe.
Fig. 4. Typical velocity distributions * This lies somewhat higher Than modern fortnulatiom for smooth
surfaces for reasons explained in ( 7 ) .
Proc Insm Mcch Engrr Vol 176 1962
REYNOLDS N W B E R , R ~ = +
I
I
turbulent friction line then represents the 'residuary'
resistance as defined by Froude w?lich, according to his
0.001
0
10'
1 I ....
107
. . .. ...
10'
. .
c,
.
WP)
. J.1
10'
law of comparison, should be the same for model and ship.
By setting up these intercepts at the corresponding ship
REYNOLDS NUMBER, I% - VUU
(I
R , we obtain the C , curve on the right which represents
the predicted resistance for the smooth ship. T o this has to
be added a factor dC, based on experience to allow for the
roughness of the ship's hull and other effects which will be
discussed in more detail later.
It is of interest to consider thi: degree of extrapolation in
this instance. Assuming length to be the criterion, this is
410116 = 25.6 times, but on the more correct basis of
Reynolds number or P x L product, this is no less than
130 times. This brings out the vital importance which skin 01
10'
' ' ''....I
10'
. ' .. ...I
10'
. ' ' ..Ll
10'
friction plays in the extrapolation of model resistance and
the need for a very accurate assessment of it. W I D 5 NUMBER, Rn
b
- VUV
Another point of interest from this figure is the fact that,
for its size, the model is appreciably more resistant than the Fig. 6. Extrapolation of model resistance
Proc Instn Mech Engrr Vol176 1962
experience or ‘correlation factor’ in predicting the resistance Schlichting formulation. The latter was developed from
of ships. These factors are obtained from the correlation flow in pipes and boundary layer theory (5) and is the basic
of model tests with carefully conducted trials on ships, friction line associated with the analysis of Nikuradse’s
and are discussed later. classical research on sand roughness.
It should be mentioned here that ship model tests The Schultz-Grunow formulation published in 1940 (14)
generally consist of two complementary parts : resistance is a development of the Prandtl-Schlichting formula (and
(or towing tests) and propulsion tests in which the model acknowledged as such by Schlichting*) in which a correction
is self-propelled. The latter give the propulsive efficiency was introduced as a result of extensive experiments on
which in conjunction with the resistance is used to estimate planes in a wind tunnel. The general level is unchanged but
the engine power of the ship. The propulsive aspects are, the slope is slightly steeper.
however, beyond the scope of this lecture. The Schoenherr line has been used very largely for ship-
model work in the United States of America. Both the
Schultz-Grunow and Prandtl-Schlichting formulations
MODERN DEVELOPMENTS IN appear to be in fairly general use in the aeronautical world
EXTRAPOLATION and the latter particularly in Great Britain.
Minimum turbulent friction line In recent years a new formulation for plane surfaces was
As already shown, the most important single factor in this developed by Hughes (16) based on comprehensive tests on
work is the skin-friction correction or basic friction line. planes in the experiment tank at Ship Division, N.P.L.
Whereas it was agreed in 1948 by the International Towing Planes of various sizes and aspect ratios were towed, the
Tank Conference (I.T.T.C.) that the empirical Froude longest being 255 ft which was towed at a speed corre-
coefficients should be replaced by a system based on the sponding to a Reynolds number as high as 2.5 x 108. He
more rational Reynolds number base, there has been much found that aspect ratio or length/breadth ratio had a signifi-
controversy as to which particular formulation to use. cant effect on the results, the longer the plane in relation to
Fig. 7 shows a plotting of the more well-known contendents its breadth the higher the resistance. Correcting for this
to which brief reference will now be made. effect, Hughes produced the formula shown for two-
Perhaps the best known in ship-model work is the dimensional flow, that is, the minimum turbulent friction
Schoenherr line published in 1932 (13) the form of which line. This line is slightly steeper than the Schoenherr line
was based on von Kirmin’s theory of turbulence. It was in way of the model range but is generally some 10 per cent
claimed that the numerical coefficient of 0.242 gave the best less in ordinate which has given rise to much disputation.
fit to experimental data on the friction of planes available The question of which line to adopt for this work was
a t that time. discussed at length at the International Towing Tank
It will be seen that the evaluation of the Schoenherr line Conference in Madrid in 1957 when agreement was finally
is for all practical purposes the same as the Prandtl- * See Schlichting’s discussioir ON.(15).
reached on a formulation called ‘the I.T.T.C. model-ship tion. The intercept C , would then be a much closer estimate
correlation line (1957)’ (17) also shown in Fig. 7. of the wave-making resistance (to be scaled up according
This line is at the Schoenherr rather than the Hughes to the law of comparison) than the ‘residuary’ resistance
level, but its slope, which is all important in this work, represented by the intercept C,.
corresponds more to the Hughes trend. It is perhaps Having made a closer assessment of that part of the model
significant that the Conference took pains not to label this resistance which can be properly scaled up to full size, there
a skin friction line and stressed that it was to be used as an now remains the question of scaling up the form effect on
interim measure for correlating ship and model results. I n skin friction represented by the intercept C,. Happily, the
other words, significancewas to be given to the slope of the work of Squire and Young (19) ( 2 0 ) and Scholz (21) sug-
line rather than to its magnitude. gests a very simple solution to this, namely, that the form
It is obviously important, however, to be sure of the effect is likely to be a constant percentage, say r, of the flat
magnitude of the turbulent friction line as well as its slope plate friction C,. This is based on the results of complex
and to throw light on this Smith and Walker (18) carried calculations made for two-dimensional aerofoil sections and
out tests on a 5-fi plane in a wind tunnel at R, up to about also for solids of revolution. Unfortunately, the more geo-
4x 107. The general level of resistance came out at the metrically complex ship form is not so amenable to such
Schoenherr rather than the Hughes level but they expressed calculation but it seems reasonable to assume that the same
the view that the tests were not sufficiently comprehensive behaviour would apply. As a matter of interest, Scholz’s
to define a new formulation. There the matter rests at results for aerofoils and Young’s results for solids of revolu-
present, but in view of the conflicting evidence one wonders tion are reproduced in Fig. 8 which shows the form effect in
whether the various experimenters are measuring the same terms of the thickness ratio d/L, where d is the maximum
thing. It is known that resistance is affected by degree of thickness or diameter of the body and L the length. These
turbulence and as this is induced artificially by various relations were shown to be independent of Reynolds
devices it could well differ from natural turbulence. Indeed number.
one may well ask how natural turbulence should be defined In the light of the above, therefore, it would seem that a
and whether in the circumstances one could expect a more correct extrapolator would be a line which is tangential
unique turbulent friction line. to the model resistance at low speeds and is a constant per-
Fig. 7 also includes a formulation by Telfer which is part centage higher than the line for two-dimensional flow as
of a system of extrapolation referred to later. shown in Fig. 6b. Proposals on these lines were made by
Hughes in 1954(16) in conjunction with his own skin-
Form effect friction formulation. It will be clear from Fig. 66 that this
A matter to which considerable attention is now being concept gives a lower prediction for the smooth ship than
given is the assumption that the skin friction of a ship form the usual process illustrated in Fig. 6a.
can be taken as that of the ‘equivalent plane’ of the same Proposals very similar to those just discussed were made
len,g.h and wetted area. Experience has, in fact, shown that by Telfer as long ago as 1927 (22). He used a basic friction
these are significantly different, the resistance of the ‘three- line or ‘extrapolator’ developed from pipe flow data of the
dimensional’ form being higher for the following reasons : following form :
(a) the augmented speed of flow around the ship form C, = 04012+0.340Rn-* . . (15)
as compared with a flat plate; Form effect as discussed above was implicit in Telfer’s
(b) a pressure resistance of viscous origin brought system and he claimed that this was taken into account by
about by the development of the frictional belt along the
length.
The pressure resistance (b) corresponds to the aero-
dynamicist’s ‘form drag’ and is additional to the wave- (SCHOLZ)
making resistance which also manifests itself in the form
of normal pressures on the hull. The total effect of (a) and
(b) is usually referred to in naval architecture as the ‘form
effect’ on skin friction and some idea of its magnitude can
be obtained by running a model at very low speeds where
the wave-making resistance is negligible.
This is illustrated in Fig. 66 where at the left-hand side
the resistance of a model has been plotted above the
Schoenherr line. It will be noted that the low-speed end of
the curve is roughly parallel to the friction line and, as the
wave-making effects are obviously negligible, the resistance
here must be essentially of viscous origin. There is little d/L
doubt, therefore, that the real level of the viscous resistance Fig. 8. Form or ‘thickness’ eflects on skin friction based
should be at XX and not at YY as under the usual assump- on calculations
Proc Instn Mech Engrs Vol 176 1962
adjusting the coefficient of R,-$ or the ‘extrapolator slope’ constant Froude number ( V / d g y ) . This presentation
to suit the particular ship form. He quoted 0.366 as an amounts to simultaneous compliance with the Reynolds
average value for ordinary merchant ships. Owing to the and Froude laws over the range covered without making
factor 0.0012 in the equation, the corresponding form effect any assumptions at all. If Froude’s law holds, however,
is not quite a constant percentage of the basic friction but is and wave-making and friction are independent (or the
fairly close to it. Very recently, Telfer changed his two- interdependence is similar in each model) the cross-curves
dimensional extrapolator slope from 0.340 to 0.300 to should be parallel and their slope related to the basic friction
bring it in line with Hughes’s data (23)and, as a result, the line or extrapolator. Telfer’s intention was that his basic
two formulations agree fairly well in way of the practical two-dimensional extrapolator should apply and that the
range of R, as shown in Fig. 7. cross-curves of constant V / t / g z should define experi-
Another system of extrapolation involving variable form mentally the extrapolator slope for the three-dimensional
effect to suit different ship forms has been proposed by ship form. To facilitate this he linearized his extrapolators
Lap (24). by plotting R,-+ as abscissa. This principle, of course,
including the linearization, can equally be applied to any
The model family technique other R,-based formulation.
Another technique for studying the extrapolation problem Whereas this process seems very attractive at first sight,
is the testing of families of geometrically similar models of experience shows that there are difficulties. T o illustrate
the same ship form in which the scale or size of the models this, reference is made to Fig. 10 which shows results for
is systematically varied. This was originally proposed by a well known set of ‘geosims’ (as Telfer calls them) which
Telfer in conjunction with his extrapolation system (22), are unique in that the measured full-scale resistance of the
and, together with his method of plotting the results, was ship is also available. This work was carried out as part of
an ingenious concept. It goes back to the fundamental the British Shipbuilding Research Association’s (B.S.R.A.)
relation investigations on the Lucy Ashton ( 2 5 ) where, perhaps for
f-
R
P V2S =+[y;a]
VL v the first time, the resistance of a full-scale ship was ac-
curately measured.
The models were all tested at the Ship Division, N.P.L.
The proposal is shown schematically in Fig. 9 where a and ranged in length from 9 to 30 ft; the ship, a former
series of model results are plotted on the same R , base Clyde paddle steamer, was 190.5 ft long. The ‘extrapolation
and the resistance curves are cross-connected by curves of diagram’ shown in Fig. 10 refers to correlation on the
I1
0.0012
0
13 2 PS 3 4 s 7 0 1.5 2 3 4 s 7 9 2 5
REYNOLDS NUMBER, R, = V L / V
Fig. 10. Correlation of ship and model results for Lucy Ashton
Schoenherr basis and for convenience the abscissa has been to a form effect of 8 per cent suggested by the general level
contracted so as to linearize this formulation. It is clear of the model results at low speeds. This gives a fairly
that all the models over 16 ft long have exaggerated resist- reasonable correlation of the corrected results, bearing in
ance due to the restricted tank boundaries or ‘blockage’ mind their limited consistency and does not suggest any
and an attempt has been made to correct the results for this significant departure from Froude’s law. The corresponding
effect using recent work by Hughes(26). It is apparent, smooth ship prediction lies somewhat lower than that
however, that the consistency of the corrected spots is not actually measured and is explained by the roughness of
such as to define accurately the extrapolator slope by the the ship’s hull, referred to later. In this connection it should
model results alone. A difficulty here is doubtless short- be mentioned that in model testing no attempt is made to
comings in the blockage correction, a really rigorous solu- simulate the structural and paint roughness present on
tion for which has yet to be found. ships and it is generally considered that the surface finish
It will also be noted that the spots for the smallest model produced is ‘hydraulically smooth’, that is, not resistance-
lie rather high at low speeds and it has been suggested that sensitive as compared with that of a perfectly smooth sur-
this might be explained by separation effects. In this con- face. In spite of the present limitations in the interpretation
nection, it is known that the tendency for flow separation of the results of model families, Fig. 10 nevertheless gives
is more pronounced the smaller the model as discussed in a good general impression of the extrapolation problem
the next section. On the other hand, some ‘geosim’ experi- and of the gradual reduction in the specific resistance in
menters have had difficulties in eliminating laminar flow ship forms with increasing size.
on their smaller models, which is another phenomenon
encouraged by small Reynolds numbers. General comment
It would appear therefore that at the present stage of It is apparent from the foregoing that considerable effort
development models can be too small, and for a given and thought have been applied over the last few decades
tank too large also. to rationalizing the extrapolation of model resistance which
These difficulties have seriously impaired the usefulness is primarily a question of skin friction. In spite of this,
of the ‘model family’ technique in throwing light on this however, it cannot be said that hality has been reached
problem and unless these effects can be satisfactorily over- either as regards the basic friction formulation or the
come we must rely very largely on the correctness of the method of dealing with form effect and, so far as the
skin-friction formulation itself in extrapolating model practising naval architect is concerned, predictions are still
resistance. made on the basis of Froude’s assumptions but without
A series of parallel cross-curves of constant Froude necessarily using his frictional coefficients. I n this con-
number has been drawn in Fig. 10 at a slope corresponding nection the I.T.T.C. 1957 line seems to have been used
Proc Insrn Mcch Engrs Vol I76 1962
very little and so far as can be ascertained the Americans for a programme of work at N.P.L. with the objective of
are still using Schoenherr and the British the Froude throwing light on these matters and the physics of ship
coefficients. resistance generally. This work has involved wind-tunnel
It is necessary in these circumstances to use correlation tests on a double or 'reflex' model of a mathematical form
factors based on experience with similar ships which are to measure the skin friction in the absence of waves and
particular to the method of extrapolation. Although satis- tank tests on a similar model from which the skin friction
factory predictions can be made in this way it is obviously in the presence of waves has been determined.
important to reduce or eliminate such empiricisms as far Some preliminary results of this investigation are shown
as possible. The outstanding matters are being actively in Fig. 11 which refers primarily to the tank experiments
pursued by the authorities concerned and in particular the on a model 20 ft long x 2 ft beam x 1.25 ft draught, having
International Towing Tank Conference. parabolic transverse sections and water-lines. The model
The fact that the I.T.T.C. 1957 line has not been taken was subjected to detailed pressure plotting and by inte-
up to any great extent has been ascribed to lack of experience grating the pressures over the surface the total pressure
of the appropriate correlation factors to be used with it. resistance was determined. The upper full line represents
In this connection it is to be stated that t h i s formulation is the total resistance measured on the dynamometer and by
empirical in that it was not intended that any form effect subtracting from this the total pressure resistance, the true
should be applied in conjunction with it-indeed, it was skin friction was obtained as shown by the lower full line.
supposed to include an average allowance for this. In the The skin friction measured on the wind-tunnel model is
author's opinion, however, a rational treatment of form also shown as are the Schoenherr and Hughes plane friction
effect must be made sooner or later which may well mean formulations.
revision of this line. It will be noted that the skin friction determined from the
wind-tunnel tests follows closely the trend shown by Hughes
INTERDEPENDENCE OF THE COMPONENTS and Schoenherr and is about 14 per cent above the Hughes
OF SHIP RESISTANCE and 3 per cent above the Schoenherr values. Including the
Yet another assumption in the application of the law of 'form drag', these percentages become 16 and 5 per cent
comparison is that the frictional and wave-making com- respectively. These latter percentages are, of course, the
ponents are independent and separable. This even applies form effect on skin friction discussed earlier and it is of
to the refinements discussed in the previous section and, interest to mention that according to Young (Fig. 8) a body
whereas there is good reason to believe that this is a reason- having this slenderness ratio would be expected to have a
able assumption, it certainly needs investigation. I n this form effect of 5 per cent, which, suggests compatibility
connection, B.S.R.A. initiated and has been responsible with the Schoenherr rather than the Hughes line.
0.007 -
TOTAL RESISTANCE
(TANK)
0.006 - SKIN FRICTION IN
ABSENCE OF WAVES
(WIND TUNNEL)-?
0.005 -
/
I TOTAL PRESSURE
I RESISTANCE
(WAVEMAKING AND
HUGHES PLATE
FRICTION
1 SKIN FRICTION IN
LPRESENCE OF WAVES
0.001 - (TANK)
01
0.4 06
0.2
0.8 1.0
0.3
1.2
.-
0.4
1.4
-2 -. 1.6
0.5
.v/w , 1 . . - . I _ L - I
An interesting and perhaps not unexpected feature is The extent of the laminar flow on the 0.75 block form
that the skin friction deduced from the tank tests is un- amounted to about the first 10 per cent of the length of the
dulatory and doubtless affected by the presence of the wave model. Since that time, however, the trend has been for
system generated by the model. I t will be noted that the the fullness of cargo ships to increase appreciably and
humps in the total resistance curve are generally associated block coefficients of the order of 0.80 are quite common-
with hollows in the skin-friction curve and vice versa. place. T o anticipate further developments along these lines
It is clear, therefore, that in this instance the skin friction N.P.L. in conjunction with B.S.R.A. has been carrying out
is not independent of the wave-making resistance but it does exploratory tests on forms as full as 0.85 block coefficient.
not follow that this necessarily affects the extrapolation to Very careful attention needs to be given to turbulence
the full scale; provided the effect is generally similar in stimulation on these models of such extreme fullness but
model and ship, it may make little difference. This needs flow detection techniques have shown that this can be
investigation, however, and similar experiments on larger achieved. T o illustrate the effect of fullness, however, it is
scale models and also on models of different form would of interest to mention that without turbulence stimulation,
throw light on the problem. laminar flow was detected as far back as 17 per cent of the
length abaft the stem on an 18-ft model of 0-85 block
coefficient.
LAMINAR FLOW AND SEPARATION EFFECTS The possibility of inducing laminar flow on ships will be
Reference has already been made to the capricious effects discussed later but, in passing, it is of interest to refer to
of laminar flow in connection with Froude’s experiments the well-rounded entrance water-lines on sailing yachts
on the friction of planes and the effect this had on his which might well be conducive to laminar flow and im-
empirical analysis of them. This matter came into promin- proved performance as a consequence.
ence again just over ten years ago when it was discovered Another flow phenomenon which may be present on
that models of the fuller mercantile forms were subject to small-scale models but not necessarily on ships, or at least
laminar flow at the fore end to a quite significant extent. to the same extent, is that of separation of flow in the after-
As there was no reason to believe that a corresponding body. The position is that as the boundary layer or frictional
laminar area existed on the ship this meant that the belt develops along the ship’s length, the layers of fluid in
phenomenon had a profound effect on the ship prediction. close proximity to the hull are gradually slowed down owing
Moreover, it was found that the extent of this laminar area to surface friction, the more so as one proceeds aft. More-
depended on the shape of the fore end, and model experi- over, in the after-body, the pressure gradient in the stream
ments involving different bow shapes probably reflected flow just outside the boundary layer is such as to aggravate
changes in the extent of the laminar area rather than this effect and in certain circumstances the boundary-layer
genuine changes in resistance so far as the ship was flow close to the ship’s surface may actually be reversed in
concerned. direction. A vortex or eddy is then formed and the stream-
In particular, it was found that the pronounced pressure line flow breaks down or ‘separates’ downstream from that
gradient associated with the rounded entrance of the fuller position. This can lead to a significant loss of energy which
models was conducive to laminar flow whereas the finer manifests itself in increased resistance and can also have a
models of about 0-65 block coefficient* and below were not serious effect on the performance of propellers and rudders.
affected by laminarity to any significant extent. This The adverse pressure gradient conducive to separation is
difficulty was eventually overcome by the fitting of turbu- more pronounced the steeper the flow lines or the fuller
lence stimulating devices at the fore end of models to ensure the form in the after-body. It is interesting to note that this
that the laminar flow was broken up. One such device and effect is quite the converse of that discussed earlier in con-
probably the best known is a ‘trip-wire’ usually about nection with the development of laminar flow at the fore
0.05 in. diameter which is fitted on the surface of the model end of models. There, the pronounced pressure gradient
square to the flow near the fore end. It was in exploratory associated with a full, well-rounded bow was directed so as
testing with such devices initiated by B.S.R.A. (12)that to accelerate the frictional belt and, indeed, had a stabilizing
attention was focused on the magnitude of this effect for effect on it, to the extent that the transition from laminar to
fuller forms and which led to the general introduction of turbulent flow tended to be delayed. In the after-body the
turbulence stimulation in ship model testing. These tests pressure gradient is reversed in direction and at the same
were made on a full form of 0.75 block coefficient and it time the frictional belt is much thicker and it is this com-
was shown that turbulence stimulation increased the resist- bination of circumstances which is conducive to instability.
ance of the model by about 10 per cent corresponding to These effects are illustrated schematically in Fig. 12.
about 15 per cent on the predicted ship resistance. This was It is important to underline that this form of separation
not the first time that this matter had been raised (27)nor, is essentially of viscous origin and certainly would not take
indeed, the first time that a trip-wire had been fitted to a place in a perfect fluid. So far as its effect on resistance is
model, but the effect in this instance was very pronounced concerned, however, it manifests itself in the form of normal
and had far-reaching effects. pressures on the hull or rather a deficit of pressure on ”le
volume of displacemenr . hull abaft the separation point, with the result that the net
* Block coeficient =
length x breadth x draught pressure resistance of viscous origin is increased. It is dear,
Proc Instn ,Mech E n p s Vol 176 1961
T O SEPARATION) T O LAMINARITY)
Fig. 12
therefore, that two of the significant parameters for this range. Flow separation was suspected and confirmed by
phenomenon are boundary-layer thickness controlled by flow visualization tests using cotton tufts as indicators.
Reynolds number and the pressure gradient which in turn Some light was shed on this problem by Kempf (28) who
depends on the geometry of the form. measured the resistance of a series of submerged solids of
This problem cropped up recently in connection with revolution in which the fullness of the run was systematically
model tests carried out for B.S.R.A. at the Ship Division, varied. These bodies were torpedo-shaped, the main portion
N.P.L. in which the design parameters of a mercantile ship being cylindrical with a hemispherical fore end and the
form were systematically varied. The particular form had Reynolds number was in the usual model range (5x 106).
a block coefficient of 0.80 and the effect of changing the The resistance increased gradually with increasing fullness,
position of the longitudinal centre of buoyancy (L.C.B.) but when the angle of run reached 18" the resistance
was being investigated, that is, moving the longitudinal increased sharply indicating that separation had taken
centroid of the immersed volume of the hull by fining the place. For a ship model, however, the surface disturbance
fore end and filling out the after end and vice versa. Some might well affect the critical condition.
of the resistance results are shown in Fig. 13 where it will An important practical consideration is whether such
be seen that in moving the L.C.B. from 1; per cent L behaviour on the model is reproduced on the full-scale ship,
forward to per cent L forward there was no appreciable bearing in mind that one of the controlling parameters is
change in resistance. When this was moved to per cent + Reynolds number. In this connection, there is little doubt
L aft, however, there was considerable increase in resistance that for a smooth hull surface the full-scale boundary layer
amounting to nearly 20 per cent in the working speed will be relatively thinner than that of the model, with the
consequence that there will be more tendency toward
separation on the model. On the other hand, the ship
boundary-layer will become thicker as the hull surface
deteriorates (due to corrosion and fouling) and it would be
prudent perhaps not to be too sanguine about this.
T o counteract this tendency in a full form it would seem
necessary to accelerate the flagging flow in the boundary
layer at the after end and the introduction of water jets
might be effective. These would need to be tangential to
the hull surface and directed aft. Another possibility would
be to suck away the 'dead' water through slots on similar
lines to analogous techniques which have already been used
in aeronautics. In this connection, the action of the ship's
propeller doubtless helps a little in stabilizing any tendency
for separation on the hull in close proximity to it.
There is little doubt that a rigorous solution of this
problem involving quantification of the critical cambina-
SPEED- knots tions of the controlling parameters is one of the more
Fig. 13. Effect of $ow separation on resistance important problems in ship resistance today, especially in
Proc Insrn Mech Engrs Vol 176 1962
view of the trend for increasing fullness in cargo carriers ascertain just what these roughness effects amount to in
and in tankers particularly. I n this connection devices such terms of ship resistance and a brief review will now be made
as the jets referred to above might well enable the fullness of relevant work already done and investigations proceeding.
of stern lines to be increased beyond present limits. This comes under two broad headings, namely, investi-
Pending clarification of these matters it would seem gations on ships, and those under more controlledconditions
prudent to ensure that flow separation does not take place in laboratories.
to any significant extent on the model scale which should
give some reassurance so far as the ship is concerned. Investigations on ships
Before leaving this topic, it is relevant to refer to rather Lucy Ashton trials. One way of overcoming the bugbear
puzzling differencesin the trial performance of ships which of propulsion scale effect is to measure directly the resistance
are reported from time to time. Reference is being made of a full-scale ship which means that there will only be two
here to successive trials on the same ship, usually under factors involved in making the shipmodel balance, namely,
slightly different conditions of trim, when considerable the extrapolation process and the hull roughness. Experi-
differences in power for the same speed, as much as 20-30 ence shows that the measurement of full-scale resistance is
per cent, have been reported. These are generallytwin-screw an ambitious and expensive project, but this was success-
ships and it seems to the author that this may be bound up fully achieved during the investigations carried out by
in some way with flow separationassociated with the bossings B.S.R.A. on the Lucy Ashton (25).
and shaft supports. It could be that the flow over such I n these experiments, this former Clyde steamer was
appendages is extremely sensitive to changes in trim. Flow propelled by aircraft jet engines, the collective measured
visualization tests on stationary models in flumes or circu- thrusts from which gave an accurate measure of the ship’s
lating water channels would appear to be a useful first resistance. This was considered to be the first time that the
approach in studying this behaviour. resistance of a full-scale ship had been accurately measured
and this method of propulsion effectively overcame difK-
T H E SKIN FRICTION OF SHIPS culties inherent in towing which had been used by previous
The basic outcome of the model test is ideally an extra- experimenters in this field. It is appropriate to mention
polated prediction of the resistance of the corresponding perhaps that to measure ship resistance one or other of
ship with a perfectly smooth surface. It is then necessary to these techniques must be used as, owing to interaction With
add to that the additional resistance caused by the rough- the hull, the thrust of a conventional propeller differs from
ness of the full-scale hull. For a new Ship this takes two the resistance ofthe ship.
forms : In these trials the effect of a number of Merent hull
(1) Structural roughness which depends on the type surface conditions was measured including :
of shell construction: for example, rivet heads and plate (i) the normal service condition with red-oxide paint;
overlaps in a riveted ship, and weld beads in a welded (ii) as (i) with the plate edges faired off;
ship. (iii) as (ii) but coated with bituminous aluminium
(2) Roughness on the body of the plates depending on paint;
the surface preparation before painting, the method of (iv) as (iii) after various periods of fouling.
application and the texture of the paints themselves. The ship resistance shown plotted in Fig. 10 is for con-
For a ship which has been in service, further roughening dition (i)-the normal service condition, and is about
takes place due to (i) breakdown of the paint system and the 16 per cent above the smooth ship prediction from the
formation of corrosion products; (ii) organic fouling such model tests. In terms of resistance coefficient, this intercept
as weed and barnacles. AC, or ‘roughness allowance’, as it is called, is about 0.0004.
It is in this field of hull roughness that much remains to As for the rest, the results are now fairly well known and
be done although, as will be seen, a considerable amount of it is not proposed to go into them in any detail. Suffice it to
information is already available. The obvious approach say there was a noticeable difference between the red oxide
would be to attempt to bridge the gap between the ‘smooth’ paint and the slightly smoother aluminium paint, amounting
model and the ‘rough’ ship by comparing trial results for to 3 per cent on total resistance and 5 per cent on the
actual ships with model predictions, but it is not easy to estimated skin friction. Fairing the plate edges reduced the
isolate the roughness resistance because of other extraneous resistance a further 3 per cent. After allowing the hull to
effects. One difficulty in particular is that we are not yet foul, gross increases in resistance of over 30 per cent were
quite certain about the smooth ship prediction for reasons measured corresponding to about 50 per cent on the skin
already discussed and there are other unknowns too, in- friction (30). The fouling here consisted of a scattering of
cluding propulsion scale effect* which arises when models +
small barnacles about in. high on the flat of the bottom
are self-propelled. It is most important, nevertheless, to together with a band of fine grasses on the side which, as
* By this is meant the extent to which the full-scale propeller (and fouling goes, could not be described as severe.
irs interaction with the hull) does not follow the scaling laws applied These tests drew particular attention at the time to the
to rhe model test. This is beyond the scope of this lecture, but for sensitivity of full-scale resistance to small roughnesses-
those interested it is suggested that reference be made to recenr work
by van Manen and Lap (29). those with the clean ship demonstrating the importance of
Proc Instn Mech Engrs Vol176 1962
obtaining the smoothest practicable hull surface in the first model propulsive efficiency applies without correction to
place, and those in the fouled conditions the importance the ship, that is, that propulsion scale effect is negligible.
from the economic point of view of keeping the hull free Alternatively, an estimate of ship resistance can be made
from marine growths. from the measured power in conjunction with the model
In an attempt to throw light on these results, very careful propulsive efficiency and a correlation made in terms of
records of hull roughness were taken with a wall gauge on resistance, which is preferred in the present context.
the two paint surfaces used. Analysis of these showed that Referring to Fig. 6 this can be expressed as follows :
the nature of the roughness was a complex function of CT(rough ship) = C,(model)+C,(smooth ship)+dC, (16)
superposed undulations, the mean amplitude depending
The correlation factor is now dCF, or the allowance in terms
on the wavelength, and there appeared to be no simple
parameter which could be used to define the surface. A of resistance coefficient necessary to complete the ship-
model balance. If for the sake of argument, we accept the
form of harmonic analysis was applied which showed that
assumptions referred to earlier regarding the extrapolation
the amplitude decreased with diminishing wavelength and
process, then the factor ACF could be ascribed to hull
that over most of the range the aluminium paint surface
roughness, and this explains why these are sometimes
was the smoother, the amplitudes being about 3 those of
referred to as ‘roughness allowances’. If we then add AC,
the red oxide which was in keeping with the measured
to the calculated smooth skin friction for the ship, we
resistances.
obtain what is, in effect, an estimate of the skin friction of
Apart from the evidence of the roughness records, the
aluminium paint surface had the appearance and touch of the rough ship.
Fig. 14 shows ship-model correlation factors in this form
a smoother and harder surface. This paint differs from red
plotted against R, for ships of two different types of shell
oxide in that the aluminium filler is lamellar rather than
construction :
granular and, on drying out, the small metallic flakes align
themselves parallel to the surface and interlock to form (1) half-welded hulls (welded butts and riveted
what is, in effect, a thin layer of aluminium foil over the seams);
surface of the hull. (2) all-welded hulls (welded butts and seams).
Hull roughness emerged from this work as a very im- In both instances CF has been determined using the
portant factor affecting the performance of ships and, as a Schoenherr friction formulation. The full lines refer to
direct consequence, some ten years ago B.S.R.A. began actual ship trials and the broken line to comparable CF
makmg hull roughness surveys on the trials of all ships values determined from the Lucy Ashton investigation. A
with which they were concerned. It was hoped that these broken line also shows the level of 3CF = 0.0004, a value
might explain puzzling differences in performance even often associated with the Schoenherr formulation.
between sister ships, which is discussed below. It will be noted that whereas there is a definite tendency
for the all-welded results to plot lower on the average there
Trials on new ships. Pending a complete solution of the is a large amount of scatter in each class. The difference in
extrapolation problem it is important from’ the practical the average levels amounts to about 7 per cent on power
point of view to correlate model results with those of care- between an all-welded and half-welded hull. The scatter
fully conducted measured-mile trials for a range of ship even for sister slips can correspond to variations in power
types and surface conditions in order to develop empirical as much as 23 per cent.
‘allowances’ or ‘correlation factors’ which can be used to It will also be noted that a number of results come below
predict trial performance. the Schoenherr line, implying that the ship friction is less
Work on these lines has been the subject of joint research than that for a smooth surface. It has to be pointed out,
between B.S.R.A. and the Ship Division, N.P.L. for over however, that this may be due to shortcomings in the
ten years, and British shipbuilders and owners have been method of extrapolation and also to propulsion scale effect
most helpful in providing the necessary facilities on the which has been neglected. Strictly speaking, the only
acceptance trials of ships and, on occasions, in arranging results in Fig. 14 which truly aspire to be ‘roughness allow-
special trials for the purpose. In this connection, B.S.R.A. ances’ are those for the Lucy Ashton which, as already ex-
is primarily responsible for the ship measurements and plained, are based purely on comparison of resistance. The
N.P.L. for the corresponding model tests. average value of the Lucy Ashton curve in Fig. 14a corre-
In order to obtain the best possible correlation between sponds to dCF = 0.0002 about, and this is a fair average
ship and model, particular attention is given to those trials of the other results taken as a whole, but one has to bear
carried out under good weather conditions, and, where in mind the possible deviations shown. The results refer
necessary, model tests are made after the trial so as to particularly to single-screw ships up to 675 ft long and
reproduce accurately the correct draught and trim. further details are given in (31).
After correcting for wind resistance, the power for a given Reverting to the scatter in Fig. 14, experience on the
speed is then compared with that predicted from the Lucy Ashton suggested that hull roughness was likely to
smooth model in the usual way and the ratio of these is be a major factor and comprehensive roughness records
usually referred to as the ship-model correlation factor. are taken on these trial ships using the B.S.R.A. wall rough-
In making this comparison it is usual to assume that the ness gauge specially developed for the purpose. These
Proc Znrtn Alech Engrs I’ol I76 I962
00030
0.0025
u) 0.0020
n
>
Q
-IN
\ 00015
b‘
u,
c, 0~0010
QOOO5
screen for harmonic analysis, as mentioned earlier, and the ‘SINGLE FOOT
‘mean apparent amplitude’ for a wavelength of 2 in. is
usually taken as a typical measure of the roughness.
Although this choice of roughness parameter is arbitrary
PROBE WITH S i n DIA.
BALL POINT CONSTRAINED
-
DIRECTION OF
TO MOVE NORMAL TO TRAVEL O F GAUGE
to some extent, it will be seen that it leads to reasonable TRACK
correlation with trial results. Fig. 15. Diagrammatic arrangement of B.S.R.A. wall
A histogram showingthe distribution of average roughness roughness gauge and track
Proc Insrn Mech Engrs VoI 176 1962
among a sample of 68 new ships is shown in Fig. 17. categories are shown in Fig. 18. The uppermost line shows
This shows a fairly normal distribution and the most fre- a record taken on a bad surface, the middle line on an
quent or average value is about 0.007 in. The roughest is average surface and the bottom line on the best hull surface
about twice and the smoothest about half the mean value measured.
respectively, and typical roughness records for these On the uppermost line the background construction to
the analysis method is also shown. This consists of a 'grid'
of lines spaced 2 in. apart along the length of the record
together with upper and lower envelopes of the roughness
profile which take in the highest and lowest peaks in each
2-in. division. The mean separation of the upper and lower
envelopes is then the 'mean apparent amplitude' for a wave-
length of 2 in. and these in turn are averaged for the entire
hull. A fuller account of this process is given in ( 2 5 ) .
More than half the sample of 68 hulls referred to in
Fig. 17 had not received any special treatment apart from
the usual wire brushing. A number of flame-cleaned hulls
were included and some which had been descaled by
pickling, but these did not appear to have any significant
effect on the distribution of roughness. Before drawing
definite conclusions, however, larger samples are required.
I n order to disentangle the separate effects of the various
factors affecting trial performance, statistical methods are
a Carriage showing probe, stylus and slide holder.
now being used at B.S.R.A. and, so far as the analysis has
In
MEAN ROUGHNESS VALUE FW
$20
SAMPLE OF 68 HULLS 00074 In. -
8 15
a
w
!$lo
3
2
5
AVER-E SURFACE
---
--
- 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I I I I I IIIIII 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
TENTHS OF AN INCH
Fig. I N . Typical records of hull roughness
Proc Insrn Mech Engn Val 176 I962
gone, plate roughness has proved by far the most significant careful analysis of service performance records, and work
factor as, indeed, was to be expected. Present indications on these lines is being carried out by B.S.R.A. on a number
are that an increase in the average roughness of 0.001 in. of ships using voyage data sent in by ships’ staff. Here
would increase the resistance of a large, new, single-screw again, there is a multiplicity of variables, however, in-
ship by about 2+ per cent. This means that we would expect cluding the effect of wind and waves on particular trade
the bad surface in Fig. 18 to require 17 per cent more power routes and statistical processes are again being used to
than the average surface and the good surface some 8 per disentangle the separate effects.
cent less, making a total difference between the two In order to isolate the effect of deterioration of hull
extremes of 25 per cent. By far the greatest part of the surface, B.S.R.A., in conjunction with Shell Tankers, has
variation in hull surface roughness is due to the uneven recently been carrying out a series of ‘repeat’ trials on two
application of paint. The figures quoted therefore underline sister ships after various periods in service, the results of
the benefit to be gained from the careful painting and which have been very revealing. These are 18000 tons dead-
preparation of the hull before trials and the penalty to be weight tankers and the procedure has been to run measured-
paid for bad workmanship. mile trials before and after routine dry dockings covering
Similar roughness surveys have also been made on periods in service up to 4+ years. On each occasion, hull
propeller blades. Whereas this is outside the scope of this surface roughness was measured in dry dock before and
paper, it is of interest to mention that for new bronze after cleaning down and repainting.
propellers the effect on performance of variations in surface The results for these ships are summarized in Fig. 19,
finish are quite small and not more than about 2 per cent. which shows the percentage increase in trial power over
After roughening in service, however, the effect may be that for the original acceptance trial of the new ships
much larger and this is being investigated. plotted on a base of time in service. In order to underline
the trends, the individual spots have been connected by
Ships in service. In the previous section reference was straight lines, but it does not necessarily follow that the
made to the significant effects which random variations in fall-off in performance with time was actually linear. In
the hull surface finish can have on the performance of new fact, analysis of service records suggests that this is initially
ships in the clean condition. After entry into service, quite rapid so that the burden of increased resistance would
however, further roughening takes place due to break-down be borne for the greater part of the time between dockings.
of the paint system, formation of corrosion products and It will be seen that there is a marked deterioration in per-
organic fouling. This all leads to increased skin friction and formance after one year and only partial recovery after dry
a consequent deterioration in performance which is of docking, cleaning and repainting. For ship By the last pair
particular interest to the shipowner. of trials after about 4; years in service, showed an increase
Some idea of this deterioration can be obtained by a in power before docking of 40 per cent, and 32 per cent after
1
u)
E
8I
L
L
W
I-
LL
4
1
ul
I
I-
z
P
%!
a
W
I-
lb
<
0 1 2 3 4 5
TIME IN SERVICE-years
Figures refer to service speed of 14) knots, still air condition and displacement of 24 282 tons.
Fig. 19. Percentage increase in trial ship over the original trial
Proc Iirrrn Mech Eiigrs Vol I76 I962
of what power might be expected for a hydraulically smooth 0 50 100 150 200 250 300 350 400
ship. This comes out at 28 per cent less than the best new AVERAGE HULL ROUGHNESS a x lo4 in.
ship trial which compares with 15 per cent obtained from Displacement 24 282 tons; speed 14i knots.
the smooth ship prediction from the model. This difference Fig. 20. Perforniance of 18 000-ton deadweight tankers.
is not surprising bearing in mind the assumptions involved Variation in trial s.h.p. with hull roughness
in Fig. 20 but it gives some idea of the scope for improve-
ment by producing better surface finish. Taking grain It is of interest to compare this behaviour with corre-
roughness as a guide Table I shows for a number of typical sponding results recently published by Professor Aerts-
ships the admissible roughness, the minuteness of which sen(33) for two cross-channel ships. Results for one of
underlines the difficulty of obtaining a 'hydraulically these, the Koningen Elisabeth, are shown in Fig. 21, in
smooth' ship. which the percentage increase in power is again presented
It is appropriate to mention here the difficulty of develop- on a base of time in service. The docking periods are clearly
ing a roughness parameter which will be significant as far evident by the sharp drop in the curves at roughly twelve-
as resistance is concerned. The nature of the roughness month intervals and the intervening curves refer to the
varies in different parts of a hull (for example, paint drops actual fall-off in performance as determined from records
tend to accumulate on the flat of the bottom) and alterna- taken in service. It will be seen that the rate of increase is
tively, the same degree of roughness may well have different highest immediately after undocking. The striking feature
effects on resistance according to the position in the ship. here is that after cleaning and repainting the original per-
The choice of mean apparent amplitude at 2 in. as the formance is very nearly regained and, after 3 years, the
roughness parameter, was largely arbitrary, and some other difference only amounts to about 3 per cent. The results for
parameter might well improve the correlation shown in the second ship were very similar. This is in marked con-
Fig. 20. trast to the behaviour of the tankers discussed earlier, one
Table I
-
I
Type of vessel Length Beam, Draughi Displace- Service Block Maximum Maximum
B.P., ment, speed coeffi- width of width of
ft tons V,, cient turbulent laminar
ft ft knots boundary boundary
layer, ft layer, in.*
-
Tug
Trawler
.
Cross-chan;lel
Dry cargo .
100
185
356
419
28.0 12.0
32.0 1543
46.6 11415
60.0 27.02
483
1390
3024
13428
I 1 11.5
12-5
22.0
13.5
0.503
0.572
0.534
0.693
1.15
0.92
1.17
0.66
1.52 X 108
3.05x 10s
1.03x 109
7.45x 10s
0.00079
0.00073
0.00042
OW068
1.04
1.75
2.86
3.51
0.48
0.62
0.65
0.90
Destroyer .
18 OOO ton deadweigh;
320 32.0 9.25 1500 30.0 0.550 1.68 1.27x 109 0~00030 2.50 0.53
tanker . 530 69.3 29.70 24311 14.5 0.782 0.63 1-01x 109 OW063 4.27 0.98
36 OOO ton deadweigh;
tanker . 660 90.0 36.38 48 733 16.0 0.790 0.62 1-39x 109 000057 5.10 1 a4
65 OOO ton deadweight
tanker . . . 775 112.5 43.0 85 500 15.0 0.800 0.54 1.53x 109 O~Ooo61 5.92 1.16
Passenger liner . . 740 97.0 31.30 37 900 27.0 0.590 0.99 2.63 x 109 0.00034 5.28 0.85
Transatlantic passenger
liner . . . 950 101.5 35.0 52 500 30.0 0.540 0.97 3.76 x 109 0.00030 6.52 0.91
-
Width of laminar boundary layer taken to the point where velocity equals 99 per cent of the main stream velocity.
Proc Instn Mech Engrs VoI 176 1962
Laboratory investigations
Whereas laboratory work on roughness effects has the
advantage of enabling investigations to be carried out under
controlled conditions without the extraneous effects en-
I I countered on ship trials, there are disadvantages also.
1 2 3 These include the difficulty of reproducing realistic ship
YEARS IN SERVICE
roughness on the test surfaces and of obtaining Reynolds
Fig. 21. Loss in performance with time in service for the numbers sufficiently high to make the results meaningful.
cross-channel skip Koningen Elisabeth Perhaps the best known investigation was that of
Nikuradse, who measured pressure drop in pipes in which
of which, in the clean condition, was 27 per cent above the the wall roughness was systematically varied by the appli-
performance of the new ship after a period of about 3 years. cation of sand grains of various diameters(36). These
In the case of the cross-channel vessel, the fall-off in results were later translated by Schlichting (8) to equivalent
performance was almost entirely due to fouling and the flat-plate friction referred to earlier in connection with
description of this (33) tallies closely with that encountered Fig. 2a. It will be remembered that a particular character-
on the Lucy Ashton, the effect on power being about the istic of this form of roughness is that for a given grain size
same in each case, namely an increase of 30 per cent. Un- the resistance coefficient rises above that for the smooth
like the tanker, however, there appeared to be little deteriora- plate and after a transition region reaches a steady value,
tion in hull surface due to corrosion and this may well be with increasing speed.
explained by the fact that the cross-channel ship was fitted This pattern of behaviour was subsequently confirmed
with cathodic protection whereas the tanker was not. from work carried out at the Ship Division, N.P.L. on
To sum up, this work has clearly shown the appreciable behalf of B.S.R.A. (37)) which involved resistance tests on
extent to which the performance of a ship can deteriorate an 1 8 4 plank the surface of which was systematically
due to increased skin friction as a result of roughening of roughened with various grades of emery powder.
the hull by either corrosion or fouling. Although the per- At one time it was proposed that all ship roughness effects
centage increases refer particularly to horsepower, the effect should be expressed in terms of a standard scale of equiva-
on fuel consumption for the same speed will be of much lent sand grain roughness, but there is evidence to suggest
the same order. It is clear, therefore, that considerable that hull roughness does not follow this pattern. I n par-
economic advantage would accrue if this surface deteriora- ticular, from tank tests on a plane treated with paints of
tion could be arrested and it is apparent that there is different roughness Todd (38) showed that the resistance
considerable scope for improvement both as regards anti- coefficient for a given roughness rose above the curve for a
fouling and anti-corrosive measures. In regard to the latter, smooth plate and then tended to align itself parallel to it,
cathodic protection seems to have been very effective on the thus continuing to decrease with speed (that is, constant
cross-channel ship. It may be that some improvement could AC,). This behaviour was also confirmed from trials on
also be brought about by more careful preparation and ships.
treatment of the shell plating, the additional cost of which The explanation might be that ship roughness is generally
would be more than offset by the improvement in ship per- milder than sand roughness which is a continuous succes-
formance. In this connection, in Professor Aertssen’s work sion of sharp excrescences. The former doubtless involves
on the cargo ship Lubumbashi (34) it was found that after some ‘waviness’ the effect of which is probably more in the
five years in service there was a power increase of some nature of a form effect than an increase in turbulence. It is
17 per cent which was attributed to deterioration of the probable, however, that the actual state of affairs is a com-
hull surface due essentially to the effects of corrosion. At promise between the two, particularly after fouling has set
the end of this period the hull was sand-blasted in dry dock in.
with the result that the original new ship performance was In this connection reference has already been made to the
very nearly regained. difficulty of developing roughness parameters which will
The treatment of new ship plates before painting has be significant so far as resistance is concerned and this is
been the subject of considerable research by the British certainly one of the more important problems in ship
Iron and Steel Research Association(35) with which resistance today. T o throw light on it B.S.R.A. sponsored
B.S.R.A. has been closely associated. This has involved fundamental work at Cambridge University in which a
controlled raft tests in which sample plates were immersed novel technique was developed for testing paint and other
in sea-water after various treatments and painting schemes roughness under boundary-layer conditions equivalent to
had been applied. It is beyond the scope of this lecture to the highest Reynolds numbers encountered on ships.
Proc Inrtn Mech Engrs Val 176 1962
I n this work air was used as the fluid medium and by Table 2. Effect of structural roughness on sht$ resistance
taking advantage of the greatly different density and vis-
cosity of air and water it was possible to obtain the appro-
priate conditions in a pipe of 2 in. diameter the walls of
which were suitably roughened. The resistance was
ship
1 I I I 1 s,e:11;1
Type of Tug Trawler Coaster Cargo Tanker Pas- Atlantic
ship lincr
determined from the pressure drop and some preliminary
results have been published by Sacks (39). Only limited
roughness excrescences could be introduced into a pipe of
this size, however, and further work on these lines is
Length,ft
Speed,
knots 1 I 1 1 1 1
110
94
147
12
1 250
12
500
15
610
16
I 680
21
965
29
0.001
I
-
I
-m 65000 tons DWT TANKER
a FRICTION (BLASIUS)
tons DWT TANKER
0
m7
-' I I 1 108 ,
10
I I I I I I I
REYNOLDS NUMBER. R , I U / V
merchant ships delivered both at home and abroad over less than 80 per cent of the installed power is expended in
the past three years based on information published in the overcoming frictional resistance. Moreover, this applies to
technical press. These are given in three forms, namely, in ships in the clean condition. If some allowance is made for
terms of (i) number of ships delivered; (ii) aggregate gross the appreciable deterioration which can set in, a figure of
tonnage; (5)aggregate installed horsepower, and dis- 85 per cent would seem to be much nearer the mark.
tinction has been made between Werent'types of ship as A further comment regarding the comparison of resist-
shown. ance curves given in Fig. 22: it will be seen that for the
In Fig. 23c the approximate proportion of skin friction cargo carriers the service speed corresponds roughly to
to total resistance at the service speed has also been plotted where the resistance curve begins to rise as the bow wave-
both on the basis of conventional flat plate friction and on making begins to develop. In other words, for this class of
the real level of viscous resistance as discussed above. ship the speed is only taken up to that point where wave-
It will be seen from each form of frequency distribution making resistance is just beginning to make itself felt which
that the volume of slow-speed cargo carriers in the speed- is in line with the preponderance of skin friction discussed
length ratio category of 0.6-0.7 outweighs by far any other above. This turning point in the resistance curve depends
and of course Fig. 23c shows that it is in this region that largely on the fullness of the form, and the speed in relation
the proportion of skin friction is highest--of the order of to the length and for slow-speed cargo carriers is given
85 per cent of the total resistance. For the present study approximately by the well-known Alexander formula :
the distribution in terms of installed horsepower is perhaps
of most interest where it will be seen that nearly 50 per
V = 21/t(l.08-Cb) . . . (17)
cent of the engine power lies in this low-speed category. Estimates show that if cargo ships are driven above this
By integrating this power distribution in conjunction with speed the cost per ton mile of cargo carried begins to rise
the ratios given in Fig. 23c it has been estimated that for sharply. This means that for a given block coefficient, Cb,
this sample of ships delivered over the past three years no the speed in knots increases with absolute size in proportion
Proc Instn Mech Engrs Val 176 I962
OIL TANKERS
DRY CARGO SHIPS
ORE CARRIERS AND SPECIAL-PURPOSE SHIPS
PASSENGER SHIPS
FISHING VESSELS
TUGS
O
’ I
’ O 1
04 0 5 0.6 0.7 0.8 0.0 1.0 1.1 1-2 1.3 1 04 (M 0.6 0.7 0.8 09 1.0 1.1 1.2 1.3 I
V/fT W/r
a C
F&. 23. Distribution of number, gross tonnage and installed horsepower of ships built during the past three years with respecr to
V/\/L where V is in knots and L is length of ship in feet
to the square root of the length. This is in accordance with to cancel a favourable gradient and for roughness, a grain
Froude’s Law of Comparison and the large ship has a speed size as little as 0.01 in. might be critical. Another significant
bonus on this account. This is in addition to the slight and unfavourable factor would be initial turbulence in the
bonus on resistance for the large ship due to reduction of sea itself, about which little was known. These factors taken
the frictional component with increasing length discussed together did not lead to any great optimism in the ensuing
earlier. discussion of Lockwood Taylor’s paper in 1950.
Since that time, however, fresh impetus has been given to
L A M I N A R FLOW O N SHIPS t h s matter by the study of fish propulsion and, in particular,
For reasons mentioned earlier, model experimenters go to the reported high performance of certain marine animals,
great pains to eliminate laminar flow on models but the particularly the dolphm. There has been much controversy
position is rather paradoxical in that, so far as ships are and speculation about this and, to throw light on the subject,
concerned, it would obviously be desirable to encourage investigations were carried out by Gawn (43) at the
laminar flow, bearing in mind the considerable reductions Admiralty experiment tank at Haslar, which included resist-
in resistance which would be obtained. Aircraft wings have ance tests on a dead dolphin. From these it was estimated
been designed for laminar flow up to Reynolds numbers as that at 114 knots (the speed of a school of dolphins reported
high as lo8. Further, as a reaction from the prominence by Admiralty observers) 1.2 horsepower would be required
given to this matter over ten years ago in connection with to propel this 7-ft long creature of 187 lb weight. This
the testing of ship models, Lockwood Taylor in 1950 raised power was of the same order, but slightly higher than bio-
the question of the laminar flow ship (42). logists would accept at the time having regard to the weight
The various factors conducive to laminar flow on the full of muscle in the animal, namely 74 lb.
scale were considered, including favourable pressure It has since been stated by Gray (44), however, that
gradients and the particular geometry of the form which speeds of over 22 mile/h have been confirmed for these
would bring these about as well as the necessary smoothness animals and this cannot be explained on the assumption of
of hull surface. I n connection with these, it was shown that turbulent flow over a rigid body of the same shape. Gray
very small tolerances were required: as regards hull form has suggested that the oscillating movements of the animal’s
it was shown that a deviation in the required lines amount- tail accelerates the water in the boundary layer which might
ing to only 0.03 in. over a length of 10 ft might be sufficient well reduce or prevent turbulence. Kramer (45) (46) is also
Proc Instn Merh Engrs Vol I76 I962
SEAMLESS HOSE
RIGID STRUCTURE
CROSS- SEC’TION
-‘
SECTION THROUGH STUBS
Dimensions in thousandths of an inch.
Fig. 24. Kramer’s compliant surface
different types of wax. The ‘degree of wetting’ was defined inferred from his results, published in 1936, that such
by the product of the surface tension of the water and differences in resistance as were measured were due merely
the cosine of the angle of contact of a droplet on the surface. to changes in the critical R , at which transition from
It was found that the value of this property for ‘water on laminar to turbulent flow took place, the transition being
glass’ was considerably different from that for ‘water on wax’ earlier and the resistance higher the less the surface was
with a varnished surface lying between the two. Lamble wetted. He expressed the view, however, that insufficient
evidence had been obtained to enable this to be stated with
certainty and this caution is justified bearing in mind the
‘mixed flow’ conditions on the plate and the fact that no
actual observations were taken of the extent of the laminar
flow. It would seem that this doubt needs clarifying and,
in connection with any further investigations, silicones
would now be an obvious choice as non-wetting agents.
Whereas the indications are that, apart from its effect on
transition, degree of wetting does not affect frictional
resistance or the adhesion of the fluid to the solid boundary,
one cannot help wondering whether there is not some
chemical or physical process which would break this
intimate bond. If some such means could be evolved to
introduce ‘slip’ at the hull surface, this would be a tre-
mendous break-through, and would appear to be more in
the province of the physical chemist than the naval architect
or engineer.
Yet another device which has been the subject of inven-
Fig. 27. ‘Supratnar’ hydrofoil craft tion is the movable boundary. For example, if one considers
Proc Instn Mech Engrr Vol 176 1962
largely a question for co-operation between the naval according to the ‘seventh power law’ as determined from experi-
architect on the one hand and the paint technologist and
corrosion engineer on the other. As for investigations into
boundary-layer control and damping, the co-operation of
ment, that is
‘=[%I
V
1’7 . . . . ,
=
-\ c,
ACKNOWLEDGEMENTS 0.279+ \
In
. a
The author expresses his thanks to the Council and Director where CF is the mean resistance coefficient at a given value of Rn.
of Research of the British Shipbuilding Research Associa- For the Prandtl-Schlichting formulation this can be expressed
explicitly in terms of Rn, namely,
tion for permission to deliver this lecture. In connection
with the preparation of material he also gratefully acknow-
ledges the assistance of members of B.S.R.A. staff and From momentum considerations it can be shown that
particularly Mr M. N. Parker, B.Eng. Thanks are also due
to the Director of the National Physical Laboratory, Ted-
dington, for permission to reproduce Figs. 11 and 13 and to Substituting for m, gives
the Saunders-Roe Division of Westland Aircraft Ltd for
the photograph of the ‘ Hovercraft’ SRNl.
This has been evaluated for a range of R,, from 105 to 1010 using
the Schoenherr line as basis and is represented by the following
interpolation formula :
APPENDIX I Substantially the same result would have been obtained had the
B O U N D A R Y LAYER R E L A T I O N S FOR P L A N E SURFACES
Prandtl-Schlichting or other similar plane friction formulation
been used as basis.
Turbulent boundary layers Formula (28) agrees closely with equation 18 up to R, about 107
after which it deviates to give somewhat greater values of S,as indeed
Schlichting gives the following relation for the sensible thichess,
it should. It has been used to calculate the widths of the turbulent
of the boundary layer 6, in turbulent flow ( 5 0 ) :
boundary layers given in Table 1 for a range of ship types. These
6
z
= 0’37Rn-l/5 . . . . . (18) can only be regarded as approximate, bearing in mind that the
formula applies particularly to smooth plane surfaces, but they
This was based on velocity distribution in the boundary layer give some indication of the order of the boundary laycr thickness.
Proc Instn Mech Engrs Vol 176 1962
The lmrinar sublayer As mentioned earlier it will be noted that the density p varies
As mentioned earlier, even in turbulent flow there is a very thin little with temperature but the kinematic viscosity Y changes quite
laminar sublayer immediately adjacent to the wall, and measure- significantly. It is of interest to evaluate the effect of these changes
ments have shown (51)that its limiting thickness is represented by on resistance.
v-
7 ’ y = 30 . . . . . (29) If pI and vt refer to the standard temperature and p and Y to any
other temperature, then it can be shown that the change in
Substituting for ZJ we have the thickness frictional resistance SRF is given by the relation
a = 30. Y . . . . 6 R F = [“(y)“‘-l] . . . .
v\;cf’/2 RF P I