24 JLJU 1978
ARCHIEF
DAVIDSON LABORATORY CASTLE POINT STATION
HOSOKEN. NEW JERSEY
Sailboat Test Technique
by
Paul G, Sperm
December 1958 Technical Memorandum No. l2Lf
(Revised April 1959) (Reviséd April 1960) (Revised October 1966)
Lab.
y.
STEVENS INSTJTUTE OF TECHNOLOGY
Technisc
e Hogeschool
INIR0DUCiION
1. The purpose of this note is to describe in some detail the procedure used at the Davidson Laboratory in tank testing sailboats in the heeled
condition,
and to explain the method by which the speed made good to windward, etc. is calculated from test results.Most of the material in the present note, and xmich additional information, also appears in earlier D.L. publications, notably:
T,M. 10 which describes the full size sailing trials of Gimcrack T.M. 16 which gives test results for the
model of Gimcrack, and
sail coefficients derived from the model and full size tests T.M. 17 which further discusses the sail coefficients
T0M. 108 which gives a complete set of calculations for heeled
and upright tests of a yacht
Note which gives Schoenherr friction coefficients, sail coefficients, solutions or velocity
triangles
etc. in convenient forTs for use in calculation.Notation
4The
oursc refers to the line along which the vese1 moves.
X = leeway ang1e, between course and ship s head
anglo between ship s head and apparent wind
(VA)(p
X)angle between conree and apparent wind (VA)
6
angle between shipvs head and truo wind
(VT)C -
angle between ship' s head and resultant wind force
on the sails
(all the above angles are measured in the horaontal plane)
¿- angle between ship's head and resultant wind force on the seils
(measured in plane perpendicu1r to mact, (neglet1ng mast ral4
s heel anglo
Sail Forces
The sail coefficients are defined by the relationships
= KR SA. VA2o SA. VA2
FR
Driving Force (lbs) parallel to course
FH
Heeling Force (lbs)
FH 0.05
Horizontal ccponant of Heeling 1oro acting at
right angles to couro
sin
Vertical component of Heeling Force
apparent wind (knots)
SA
Sail area (squimre feet)
-
*
Sail AÑa
3/
of Fore TriAangle
(+100
0,
of
.tinsail
J+100 /o of
J
exclUding Roach
(-*
in the sail-area notation used. in Rating Rules
b xp
).
The Centre of Effort (CeE.) is taken as the centre of
the foUdng
areas
1700/O
of Foretx-iangle at Centre of Area of Foro Triangle
100°/o of
ainsai1 at its Centre of Area
Balance predic*ions Lising this C .E. are bnd to be in reasonablz agreeìt with
hein angles observed on fuilsize yachts.
Confirmation is provided by Warners
pressure measurements on sails(Trns. SNANE l92) which showed pressures
on jib about double those on mainsail.
.
The total effect of wind on sails and hull is taken
as the reau.Ltant
of
F
FH cos Q
and
sin Q
acting through the C .E.
peeled Test!
- Applied Force System
,
The rol1owin
syst
oÍ» forces
nd couples is applied to the model
when it ïs towed in the tanks
a.
Resistance
(rt)
This force, acting f oriard alo
the course
is applied at
deck level near the bow1
by the tow-bar, which is adjusted
to be horizontal and parallel
to the course0
h.
Lateral Porco
(FH cos
or F)
Two horizontal forces, at right angles to the course,
are
applied within the hull,
near the W.L, one near each end of the nadel.
The dynamometers
are arranged with a simple mechanical
linkage, so that
a aingle moveable weight balances these
two forces.
Trie value of this wc.ght is ecual to the sum
of the two force3,
i.e
,to the total transverse
horizontal
force
('FRQ) .
The position of the weight indicates
where
cos
icts i.e0. the C.L0R. position
i'.. cos Q
is, of course, oqud
to the horizontal
component of the sail
fcrce0
C.
Vertical Fo
e - Added Weiht
(F
sin Q )
(See Figure 1)
Weights are added within
the hull at the mid section to
repre-sont th
vertical component of
the sail force
The Resultant
of i-. cos Q
and
Fsin Q
is the heeling force
F..*
Tt
should be noted that the C.L.R. position referred to
in this
note has a dffferent
meaning from the C.L.R.
discussed in tcxts
Sliding Weight Shift (S) (See Figure 1)
There is a sliding weight of two lb., which is adjusted
transversely to
get the desired heel angle. The sliding weight is initially Betifl the position which gives zero
heel angleat
rest and the sliding weight shift (3) is measured from this position to the positionwhich gives
the desired heel angle in any particular running conditioñ.
Rig Shift (See
Figure 1)
The sliding weight and its mounting (total weight 2.LO lb.) can also be moved fore and aft. It is
moved forward from
the position which gives correct trim at rest at zero heel angle by a distance known as the 'Rig Shift' which provideaa couple of appro]cLmately
the value necessary to allow forthe fact that R is aDpl'ied at deck level, not at the Centre
of Effort of the rig.
On the assumption that slight variationz in fore and aft trim are not 1mportant, standard values of rig shift
are used,, which are selected according to the type of model (Racing or Cnising) and the heel angle.
The correct rig shift, of course,
depends also on the model
resistance and
height of C.F. The¿ justification for using a fixed rig shift l'or a given heel angle
¿
is that the models are all of about the saine size, and thereforehave roughly
the same resistance at thespeed corresponding to a given heel angle.
If it were ever desired to have more exactly the correct trim, it would not be difficult to move a ballast weight fore-and-aft to provide a coutle closely
approximating the correct value. The correct value for this couple is such that
as applied, combined with this
couple, is equivaìent to
acting at the C .E. combined with
R_R8m applied at the level
where
the excess akin friction acts (i.e. somewherebelow W.L.)
(R
Fiiflsize Resistance
x
Node_DisPlacement)
8m
bhip Displacement
However, it must not be forgotten that
there is no exact data as to the height of C.E.,
for real sails.
i
i':-1214-h-Heeled Tests - Tank Procedure
The sailboat test technique
used at D.L. is based on the fact that
the model can be run down trie tank
every two ninutes or leas; (ìt is considered desirable to maintain a two minute interval between rune toavoid changes in turbulence).
Therefore, several runs are ide anu the
apparatus is adjusted eetween runs unti! the mode], runs in the desired condition.
Before running, suitable values of Heel Angle (Q , speed
(Vm) and Lateral
Force
(FNcosQ)
are
selected, and Rig Shift and Added Weight are adjusted to the appropriate values. Then runs are madevarying the
Sliding weight shift, CLR position, and Leeway angle ) , until the model runs at the desired heel angle with the desired lateral force.For the condition so obtained the following are recorded: Q
,
F cosQ
, CLR position, Sliding weight shift, X, and
At each speed it is usual to run with three different lateral forces. At each heel angle, it is usual to run three speeds.
For a
coiiplete test three heel angles, 100, 200 and 300 are usuallyrun.
This involves 9 points at each
heel angle and 27 points in all.Sometimes it is found that extra points may be required; sometimes points can be omitted. The object is to cover the
range in
which thebest speed made good to windward (V)
will lie.
Ecperience gives avery good
guide to the
values ofVm
which should be used.
It will be noted that running at chosen values of Heel Angie () and Lateral Force (FN coQ) makes it possible to fix appropriate valuez of added weight (FR sinQ) .
Furthermore, the subsequent calculations are
very much simplified by use of fixed values el'iieel
Angle. The equipnent is so designed that tests may aleo conveniently be made at heel angles ofO
150, 2S°, and 3°.
T14-12
-heeled Tests - Calculation Prcedurt
9
In testing,
FF coso
and
sinQ
are applied within the bulle.
whereas for the ship
FH co5
and FH sins are applied at the
Centre
of Effort of the eail plan.
Furthermore, the C.G. of the model is not
in 'eneral
in the sarx
position as the C.G. of the ship.
However, in the model test a heeling couple is
applIed y the shift
of the slIding weight, which is adjusted until the model
rune at the selec
-ted heel angle
lO
Given the heel angle
(Q) ,the C .G. positions of moael and
ship, and
the G.E0 posItion for the ship,
it is not difficult to calculate,
for any
value of
Fcoser,
what couple nrì$t be combined
with all the transverse
gravity and dyràmoreter forces
acting on the model to give
a resultant in
the seme line as the resultant of the gravity and sail
rorces applied to
the ship. In the teSt c couple is provided
by raovin
the sliding weight
and we thus arrive at a relationship between sliding
weight shift () and
cosG
which must be satìsf led
if the model forces
re to have the s.
line of actn as the
ship f orces.
The plot 01 this relationship
is a
straight line and is
referrod to asthe stability
line (or stability curve)
We aleo have values of
COSQand
obtained in the test
runs
and these are plotted
on the san* a
as tbe stability lino
Then
the point of interscction
gives the value of
FtT Co5
with which the ship
would rim at
the selected heel
angle.
This is referred
aa the
equi-libriun value,
In theory, different curves relating the values of
FH cos
and
S
observed in tests at
a given heel angle might be expected Lcr
different speeds.
In fact, the difference is not usually not
ieekble for
the snail
¡'ange of speed ordinarily investigated, end
so a single value of
FE cos
is ordinarily obtained
for each heel angle
T '-l?1
.6
-TM-l2I
IL The values of ail
the other quantitiee observed aro plotted against FR cosQ for each speed and heel angle The equilibrium value of FR having boon dterminod as above, the corrosponding values of the otherquantities are piked off from the curves for uso in subsequent calculations. This plotting process tonds to draw attention to any readings which aro
inconsistant with the others, and oy be in error.
12 It dl1 be noted that the equilibrium condition obtained as above, involvo the righting couple provided by
the hufl in the actual running
condition. Any chnngo in stability due to wivaiaking, etc. is thus auto-matically taken into account.It may also be noted that if it is desired to dotermino the effect on perforrrrìce of a snail change in stability, due to alterationc to the bal last or sail plan for example, it is necessary to redraw the stability line and repeat subsequent calculations but not to do any additional toet-ing, provided that the intersection still lies within the range of FR co tested, which cìunonly is the case.
13. The equilft'iuxa raluee of FR co and rust nr be scaled up
to full sie
FR cos is assumed to be unaffected by rernolds Number, c, and is therefore scaled up in the ratio of the disp].acoraente of ship and model. must be corrected for Peynolds Number and for drag of the sand-strip used to stimulate turbulence.
The sandetrip extends from the f ortmurd end of the ter1ino to the point at which the bottc of the keel bocomo approximately horionta1.
Heeled tests are run with sandstrip 1" wide (i.e. 1/2" on each ido of centreline), Upright tests are run with 1/2" and 1" sandstrips. It is acsumed that the added resistance dde to the sandetrip is proportional to ita width and that the added resisthnco is the carne for heeled and up-right tests. Furtheraoro, it is found that tho incroase of resist&nco coefficient resulting from increasing
sandstrip froro l/2
to 1" is to a
first approximation independent of speed. An average ovar the spood rangois thoref oro takon.. The ìncretise
of resistance coefficientdue to a
1" sand-strip is do±lo the difference 'cetween 1" and 1/2" and is of the order of ,COO.After tiis correction has been applied the ndal resistance is exnanded full iize using the li,e. Owing to the usually
i e.
a -
-.i 'case
(3 ô4k
CLÂht.
j
frbJ
c "'
6This idea behind the above assumption is roughly that the effect Of trimming the sails is approximately the same as the effect of rotating the whole rig about the mast so that its angle to the apparent wind re-mains the same as it was in
the Gimrack
sailing tests.On this basis, for any heel angle, a range of values of KR and cosQ maybe calculated corresponding to d1?'ferent values of + X These calculations and results appear in TM 17,
and
Figure 17-3 prèvides the data in a convenient form for use. Figure 17-3 is reproduced as Figure 2 of this note.It will be noted that in this fi: e an ä.ngle
defining
the trim of the sail ( ô) , appears as the independent variable. However, this variable does not appear elsewhere in the usual calculations and may,: for most purposes, be disregarded.The
conditions of
the original Gimòrack, sailing trials are representedby ô-2O<
and the vues of KR , and KR/KS cosQ for ô 200 are thesame as the values obtained in the trials. Th Values of ( + ?..) given
for ô - 200 are 10 larger than the values estimated for Gimcrack, since it was considered that the values for Gimcrack were rather lower than would apply general1y.
TM-121
'f
cut-away profile shape of sailing yachts, a length equal to 0.7 LWL is used in calculating Reynolds Niunber for
model and ship.
llj. Comparison of ests of Gimcrack sailed by a good helmsman with tank
results for
her model gave the values
for KR K , , and X for ¡2' Gimcrack sailing at any value of heel angle () ,However, it is not necessarily true that the value of at which Gimcrack sailed is optimum for all boats To overcome this difficulty the assumptions are made that, for a given
heel
angle (Q) , if is changed by ¿$3 , (with a suitable adjustment of sail trin)a) The resultant of KR and K1 is unchanged,
& b) The angle whìch this resultant
makes
with the centre-plane of the yacht changes by1.
Values of
R arAd F cooG,for
era3. speeds at taL heel aïg1e
being asiuilabls frcì Uia ìiio&1 testh
Ui closehaL1d p rfor.n
is
calculated as
follows;-Rs K
&
The ratio
,
i
used to ontor
Figure 17-3 and corresponding values of
and
aro read off,
b.
V,is obtained froe the oquation
R,
o
The velocity triangles now give true wind volootty
(VT),
Angle of true ind to ehtpe head (')
Speed made good to windward
Vì,gThe above calculations are nude for several speeds at each heel
angle and the reult are plotted as
Vm(ordirate) agint
(abscissa)
The
nvelope touching those
curvos gives the opthrium
ele eebau.lsd perfonnc e
o
50
+
-Vrv
Sorietios ari extra speed znay bava to be run to obte.in the part of
the fixed
Qcurvo '$iich touches the envelo, but oporionco gives a
ery good guide to the poods
id.ch should be run at each. heel anglo0
16,
L icH cosQdoes not change iuch whon
(j3 + X)
is vur'ied, tho
rition of
for a fid hoel ançle is
ul1 within the rance
con-adored.
Thus it is true that the envelope touches the fxod Q
curves
va'y nearly at the na. value of V, obtained at fixed
GThis fact is oorncti os used as a short cut to save p1ottin without
appreciably affecting the rosults
(This is dono in T
105)
TM-1
Tri- 1214
- 10
-For routine computing certain charts are used to save labour. These give:
KR and ( + A) for all values of KR/KR cos within the
range used, for = 0, 5, 10, 20 and 30 degrees
Vmg/Vs and VT/vs for all values of VA/VS and ( + A)
within the usual range. (These charts are ¡n Note 514.)
An estimate of weather or lee helm is obtained from a graphical plot based on the hydrodynarnic forces determined from the model tests and the sail force acting at the CE defined ¡n section 14 (p.1. above). For this procedure see TM 108. The balance predictions so obtained are ¡n reason-able agreement with helm angles observed on fulls ize yachts.
ADDENDUM TO TM 12k
October 1966
Numerical Calculation of Sailboat Performance
There is an alternative to the graphical procedure for performance calculations which is particularly suitable for use with a digital computer and may be convenient for other purposes.
It follows from section 114, and ¡s implicit in DL TM 17, Appendix A, that for fixed heel angle
JKR3 + KH3 = constant = KT and
+ A - tan1 KR/KH = constant = E , say
The Gimcrack sail coefficients can be concisely stated by the following tabulation of K.1. and E , taken from TM 17:
r
RH
,V
Values interpolated from smooth curves
In practice it was found that the graphical presentation of the sail coefficients given in TM 17, Fig. 17-3, was too cumbersome for routine use, and the information was re-worked into a nomograrn given in DL Note 54. Values of + À , K., and Rs/FN Gos e were taken from Note 54 and worked backwards to give values of
Kr and E The latter
agree exactly with the above tabulation. Thus the riomogram ¡n Note 5k and the above table of
Kr and E are exactly equivalent and the use
of either the nomogram (which has been in routine use for years) or the above table will result ¡n identical performance prediction.
The model tests give values of Rs and FN cos e for several speeds at each heel angle, where of course FN = (FN cos 8) sec 8
For the given heel angle, K1. and E are obtained from the above table. The relative wind angle is found from
' À = E + tan1 RS/FN
and the apparent wind speed, VA , from
Kr SA. VAa = \J R + FH 1M-124 I I -e , deg Kr X 10 E , deg 0 6.23 10.16 5*
514
9.70 10 4.77 9.29 15* 4.20 9.15 20 3.70 9.3025*
3.24 9.80 30 2.80 10.67 35* 2.40 12.00Knowing + A. , VA and V from model tests, the wind speed, VT , wind angle, 'y , and speed made good, VMG , may be found as follows
VAcos(P+A)_Vs
VA sin (ø + À)
tan' b/e
b cosec y
V5 cosy
A Fortran computer program based on this procedure has been developed at the Davidson Laboratory. The program will either take model values and expand them to full-size or accept full-size values of R5
, FH cos 9
and V and compute S + X
, VA , VMG , V1 y , etc. This program
also has provision for computirrg the balance predictions mentioned ¡n section 18.
The sanie method has been used, with the aid of a digital computer, to prepare a table, for heel angles of 10°, 20° and 30°, in which
VT/VS , S + X and y are tabulated as funGt ions of two
arguments:
cos O/R and R5/V52 . SA
The table covers the range of the arguments usually encountered, and a fairly simple double interpolation gives VMG , VT , etc., for the
actual values of the argument obtained from the tests.
TM- 124 - 12 -(see TM 108, P. 44): e = b = y = VT = VMG =
Sling Woi)it
xiol at Rost
Trimsd to O - 0°
Mded voight, applied at midsection,
aecouiit& for
daiwnrcI coiiiponnt of
rig forces.
LIi)I
;rEIGHT 3IiIFTliing Woiht
r'
Shift, in.
s,'
Added Weight
!odol Underiay
Tinnd to required heel angle by
shift of sUding weight arid by
Lyth'od,yiiainic Lift of koel.
As sail force, increace, entire
sliding uiht ri
iust be
hií'ted foririì to account for
the forrard rotational
omont
of tí sail forcuso The rig is
shifted tho distanccs given in
t.0 tai)lC bilow.
of extrcme
weight
FJ.g Shifts For
O
Racin- Liol
Cruising ?dels
k
Due to the lot: stability
racing models, the slU.ing
-\
0,,,
j
Lu..o.6l
l.&DT
l.O2
)
'['tI
4. '-.?
h-I
i
7"ri
is zlung
e1ow decks in these
casurencnts arc rac from O - 0° position,
classes,
in cruising models the
cliWLnig weight is above deck8.
--- ---h
. .E II
s . P..' -t i . _, -., çi
r
%,._. I Ai (J .j: C. O 'Ñ Q) (::-"u
M : 1 ; .1-I ' T t Il. i'I:J.:
' :'i::L-;3
,À ):\
: I I 4]1 L.'
i;î
:. .. -I -t I-: I t .; t i_ _1_:::ri
_i.J
'oLj
: . .cLI 3èinJ
-' , . : ' -1zì:.wL. I I ::;
-. ;-LÇ05
V _ ; i j 1-(i :! ---i -:-Li;s-ji:,
tIjiÌ
IPrediction of Ship Resistance from Model
The total resistance of Ship (R) and model (r) is made up of
two parts, one due to friction, the other due mainly to wavemaking but including also various smaller effects such as eddy-making. The former is caLled the skin friction resistance, (rf , Rf) or frictional re-sistance; the latter the residual resistance. (r R )
r r
i.e., or the model r = r + r
For the ship
P=R +R
f r
Ft can be shown that a model tested at speed y will have the
Rp.
same wave pattern as a ship tested at y and the value of will
then he the same for model and ship.
or in other words, if the value of -y--- is the sanie for both model and
Rr
ship, then -s-- is the same for both.
Having calculated the model frictional resistance, (rf)
, as
overleaf, we subtract this from the total resistance measured in the tank (r) to get the model residual resistance (r).
The ship re-sidual resistance is ohtaned by scaling up the model resistance in pro-portion to the ratio of the displacements of ship and model. To this is added the calculated frictional resistance (Rf) to arrive at the ship resistance
The ship resistance so obtained is for a ship speed equal to the model speed multiplied by the square root of the scale ratio.
TM -124 15 -Thus ff we have Ship Speed y = [Scale Ratio Model Speed R r
Aship
A model rwhere
The skin friction resistances may be expressed as:
and 1 2 r
C .p y
A f fm 2 m m 1 2 RC .p V
A f fs 2 s swhere and are skin friction coefficients appropriate to model and ship
density of water in test tank (fresh)
p - density of water around ship (usually salt)
A and A are wetted areas of model and ship
n s
Re P =
p = density of the fluid = viscosity of the fluid v= 1t/p = kthematic viscosity
for model
for ship
Skin friction resistance is not measured in a tank test; it is computed from tabulated skin friction coefficients, as follows:
lE the flow is turbulent, which may be ensured by sand strips on the model, and is more-or-less inevitable for a hull of the size and speed of the fuilsize yacht, there is a known relationship between and the Reynolds Number.
X is the length of the wetted surface in the direction
of flow. or normal displacement vessels X is usually taken as equal to the LWL.,
From the tabulated density and viscosity of fresh and salt water, the Reynolds numbers for model and ship at each speed tested are readily
The Reynolds number used in skin friction calculations is defined as
but for sailboats X is taken as 0.7 x LWL, owing to their cut-away form.
calculated. Then the values of for model and ship are obtained from tables of Schoenherr or other friction coefficients. (c is rather sensitive to changes of water temperature.)
A convenient procedure for carrying out the calculations is as follows:
Measure model resistance r at speed y
Evaluate C
tm
where C
=r.x
2tm 2
Apv
mm
(If a sandstrip is used to stimulate turbulence Ct is reduced by a small amount (about .0005) to allow for the extra resistance of the sandstrip.)
Determine Reynolds number for model at y and ship at V (R and R )
em es
Look up C and C for values of Reynolds number R and R
fm fs em es since
C =C
-C
f fm fs C C-C
ts tm f ASS
Resistance of Ship = R = C x ts 2This resistance is for ship at speed V v
Repeat for other speeds.
This calculation isidentical in principle to that described earlier,
and therefore A 2 V L s L and
APV2
ship s s L - 2 model A p ymm
TM-124 -17
-DA(IT.A14 5'HÖW(, FÖR(ES
CON4T?uTIN& TO SAWBOT SrALJT>1
)N H1iE
C(jNtflTON
G\
j//
Vws
f(
')t_._)_/ .1WI-\
UTÂY VTE
MCrEL OR SHIP
Ncte.
r+tly Hì
v;ûceI VCG-
c± G
'v'Hic uprrqhT c
d+«v, with
1i
s!ict,
u..rcs shcu.'r.
ur
sìY
S
p(,+(fl
r'
±;
VC& fo Cr
kflCI COUSIYK1 c
hccrI
ftc. Yvìt9
df'br
TM-12t
-TM-i 24 A-1
APPENDIX
STABILITY CALCULATIONS
Appendix A has
been added to provide additional information as to the derivatives of the "Stability Linest' outlined on pages 6 and 7, paragraphs 9-12of this
technical memorandum.Bu t
FHx2 W(b sìnQ + c cosQ)
2.00 x S
C
STABILITY CALCULATIONS
Refer to the figure on page 16 which shows the geometry of the ship reduced to model size, and that of the model itself. Because the over-turning moment of the model is less than that of the ship by the amount F1xa , the righting moment nst he correspondingly reduced by shifting
G laterally. Thus, to produce the sanie static equilibrium,
where all factors are for model size. Rearranging this expression,
FHcosQ = (b sin9 cos9 + c cos2Q)
where 2.00 = sliding weight
S = sliding weight shift
Wb . 2.00 S cos79
FHCOS
= - smn9 csQ +
a aThis equation establishes the necessary relation between FHcosQ and S
to make the model forces equivalent to the ship forces. The experimental data establish the actual relation between FHCOS and S for the model
in question, as governed by its form stability.
The intersection
of
the two straight lines which represent these two relationships gives the result sought for, namely, a valueof
FHCOS consistent both with the actual stability of the form and with thedimen-sions
of
the ship.In practice, the lines representing the two relationships lie approxi mately at right angles. Thus, their intersection can be determined graphi cally with considerable accuracy.
TM-121+
STABILITY CALCULATIONS
Rearranging terms,
APPENDIX B (March 1968)
Refer to the figure on page B2 which shows the geometry of the ship reduced to model size, and that of the model itself. Moments are taken about the intersection of LWL and centerline plane of the model.
Heeling Moment Righting Moment CE x H secp
K +(G-G) Am smp
where all factors are for model size and
G -G is distance between model vertical center of gravity and
s m
ship center of gravity location
K ¡s moment (inch-lb) measured by the test apparatus and represents the righting moment due to the water forces.
(G-G)
K + s m sirip H CE secp CE m seccp (G -G ) A sinp cospKcosp
s m m H CE + CEThis equation establishes the necessary relation between H and K to make the model forces equivalent to the ship forces. The experi-mental data establish the actual relation between H and K for the model ¡n question, as governed by its form stability.
The intersection of the two straight lines which represent those two relationships gives the result sought for, namely, a value of H
consistent both with the actual stability of the form and with the di-mensions of the ship.
In practice, the lines representing the two relationships lie
approximately at right angles. Thus, their intersection can be determined graphically with considerable accuracy.
TM-
DIAGRAM SHOWING FORCES
CONTRIBUTING TO SAILBT STABILITY IN HEELED CONDITION A m WATER FORCE TM-124 32