16 NOV. 1976
ARCHIEF
I
w
UNIVERSITY
OF
SOUTHAMPTON
department of
aeronautics
and astronautics
S.TJ.Y.R. Report No. 20
THE TACTICAL IMPLICATIONS OF THEPOLAR CURVE OF YACHT PERFOR}IANCE
P.V. NacKinnon
ADVISORY COMMITTEE FOR YACHT RESEARCH
Lab. y. Scheepsbouwkunde
Technische HogeschooI
Deift
S.U.Y.R Report No. 20.
The Tactical Impliòations of the Polar Curve of Yacht Performance
by
P.V. MacKinnon.
January, 1962 Reissued, 1967.
Summary
S.U.Y.R. Report No. 9, which coritaiis
the
papers presented at the Conference on Yacht Design and Research held atSouLhamptn University in 1962, has been out of print for some time. Most of the material in this reprrt is now available
in other pub ications, but this does rot appear to be true of the paper presented by P.V. acKmnnon, entitled 'A Scientific Approach to Yacht Racínci'.
The present report is a reprint of this paper. It discusses the tactical implications, for sailing in winds of varying strength and direction and in tidal streams, of the knowledge of yacht
rerforifiance which is repi;unted by the polar curve relating the yacht's speed (V) to the true wind angla (ì).
A recent addendum comments on the application of the saíne principles to modern catamarans.
A second addendum gives estimated polar curves for a 12-metre yacht.
A Scientific Approach to
Yacht Racinq
I shall not attempt to discuss the
tactical problems which arise from
the
presence on the course ofother co:petitors. My approach is that, for a given course in given
conditions of wind and tide, the problem is to discover the
quickest way to sail from each mark
to
the next and so to finishin tbe
shortest possible time. For considerationof this problem it. Is necessary first to know the maxirTum
performance
of the
yacht,
and this «111 dIffer for different wind velocities.Polar Curv
Let
us therefore suppose
that know our yacht's maximum speed, sailing at any given angle to the
true wind direction, and over a range of true wind speeds.
This information 1 most
conveniently shown by means of a series of polar
curves.
Each curve will apply for one windspeed only and will depict graphically the performance
of the yacht when sailed to best advantage. The distance of
the curve from
the r1gin in any given direction represents the maximumspeed of which the yacht is capable In that direction.
There will, of course, be
gap in the curve directly upwind, and
without attempting
to be preciseabout
exact speeds, the polar curve for anormal
yacht in a moderate wind wil.is the origin and the curvo is so
drawn
that any radius vector V from origin tocurve
represents the yacht's maximum speedin that direction
?
to the givenwind.
If the curve has either a hollow or a gap, thon an external tangent is
drawn
and the limit of performance to windward is the tangent. It will, however, benecessary to sail on
two
tacks to achieve lt. In Fig. 2 thépolar curve
is the saíne as in Fig. 1,but
the windward tangent hasbeen
drawn and the yacht's performanceIs limited by it.
That
is to say, thebest speed
dead to'windward Is represented by the vectors OB. But to sail In any given direction in the
windward
section, lt is necessary to maketwo
or more tacks, each of which is in aithethe direction or the direction OB.
The polar
curve
is the simplestway of setting out full information on the performance of the yacht.
A complete set of measurements has, so far as I know, nover been made,
it It is nevertheless possible to give a demonstration of how these curvos can be used (and how they will doubtless be used when
they have
been
measured). For a normal keelyacht they are likely to have the following
features-Except possibly in the downwind quadrant, the curve is convex everywhere.
The velocity Is greatest in the réaching direciion (approximately across wind).
The
two
points where maximumspeed
is made good to windward are at the highest points of curves whichare
smooth and continuous.
Even if nothing more is known about the curves than this, some interesting and valuable
deductions
can be made.I propose to show how a number of special cases In the prcblem of sailing in variable winds can be dealt with. It is a common experience that a change of wind direction results in a yacht
previously out of the running being elevated to first place. Such a happening Is not mere luck if the helmsman (a) foresaw that as a matter of geometry he would be favoured if the change of wind
direction took
place, (b) had some reason,
whether from meteorological knowledge or justa "hunch", to think that lt might take place.
My object Is to draw attention to the methods which can be used, given the yacht's polar curve, to examine the effect of possible future changes in wind direction or strength, and
so to
assess what will happen if these changes take place. The geometry Is not always simple and the results not entirely asone might
expect, so that what can be learnt from a few examples is helpful towards an understanding of general principles.Sail jnq to Windward
When faced with the problem of sailing a windward leg in a variable wind, the
helmsman
cannot of course know what is best to do unless he can foresee all the future behaviour of the wind für the period of time Involved. It wotìd then be possible in theory to select the optimum path. In practice all that can be done is tork to a few general principles,
and an examination of .certain special cases throws
light
on these principies.-3-Case i
In an
otherwise steady wind, there Is achange
of direction for a short period, after which the wind reverts to its original direction. In Fig. 3, 0 represents the point reached when thewind
changes direction; M is the windward mark. The figure is drawn for the case where thetemporary
change Is aciockwi.se.one,
and we draw from Oas origin a scaled down copy of the
polar curve whichindicates the
limit
of distance attainable in any chosen direction during the period.for which thewind
remains changed. On this limiting curve it is obvious thatP is the
point from which, in the normal wind, M can be reached in the shortest'possible time.
If the polar curve conforms to feature (jij) of those quoted above as applying generally, P does not coincide with A.Thus we sec that during the
wind shift,
ndt only is It Importantto be on one particuiar
tack, but it actually pays to sail on that tack wide of the normal angle used inwindward work.
There is also the obvious limitation that the resultonly
applies when thecourse from P
to v1is within the
windward quadrant and so cannot besailed with advantage
on one tack.
Case 2Shortly before reaching the windward mark, after a long windward leg, the wind changes direction and the change
persists
fora ionge± time than is needed to rach the
mark.
What is
the most advantageous position to be in wt-in the change takes place? Inin the given time. This line is of course limited in length but in the nature of the problem the limits do not
enter Into the
result.
A polar diagram reversed is
drawn
wIth the windward mark tas origin, and
for
the new winddirection.
This
represents the locus of poInts from which M can be reached in equal times, arid if the scale is such that it touches XY, then the point of contact is th solution to the problem.This time the solution sho that after the change of wind
direction it
pays
to sail the final leg to the mark on oneparticular
tack, and
wide of the normalangle for windward work.
Case
A windward
leg is to be sailed from the startingposition M1
in Fig. 5
to thew1ndrd mark M2
and after the lapse of approxirndtelyhalf the t.me
required
the wind changes direction and the change is permanent. In Flg 5 the polar curve H1 A1 B1 is drawn with M1 as origin to such a scale that it represents the locus of points which can hereached from
within
the time which will elapse before the wind changes. Another polarcurve reversed is drawn from M2 as
origin, and for the new wind direction.This represents the locus
of points from which M2 can be reachedin equal times, arid if
the scalo ofthe second
curve is such that It touches the first, then the point of contact is the solution to the problem.Fig. 5 shows that in sorne cases, depending
on the relative
positions
of M1and M2, and on-the amount of, and time
of, the changethe normal
angle forwindward work.
-1aving exarnnod three particular cases, I will ' tö introduce some generality. Let us consider a
windward leg to be sailed in a
wind which chances direction frequently and at random,but having a
mean direction such that the leg is always to be considered awindward one.
From the starting point, for the wind direction prevailing
at the
start, a small polar curvo
can be drawn for
points accessible within the interval of time for which the wind is expeted to remain the s arno.Let us now
forecast (or sornetrow have knowledge of)the next wind direction. From each of the points on the firt curve a
polar can be drawn for points
accessible within the nextinterval of
time.
These polars will have
anenvelope
which is the locus
of
points accessible within the second interval oftime.
So, if we cn forecast the wind changes, ? step-by-step process is possible which will in the end colve the problem of how quickly,
und by
what route,the
windward end ofthe leg will
be reached. The process just. described
is exactly anlogous to that in current use for
optimising ocean
steamship routes.Forecasts are made of wind and sea for the duration of the voyage, and from these
forecasts are deduced the shlp's
possible
speed on each day ofthe
voyage and at each point
in
the ocean.The process is, of course, of such
ctrnplication that it
can never be carried it on board a
yacht;
hut a
race is won 'the halrnsrnan who
makes thenearest
approach
o t.he correct
solution. It Is therefore necessary forand such
principles cari he best evolved by the methods used to illustrate Cases 1, 2 and 3 rather than by the method of successive envelopes. It is, for examples a ;.ìrnplc extension of Case 3to
deduce that if
the
wind is fluctuating indrectn about a
fixed average, then whenever the wind has shifted clockviso ofthe average
the yacit should be on the starboard tack,and moreover should
besailing wide of the
normal windward angle to the wind as lt is atthe moment.
It will be noticed that Cases 1, 2 and 3
are all
examples of thu use cf two polar curves, one drawn with the startthq point as originand the other drawn reversed with the
goal
as origin.right hand half of it is the line C1 B1 K1. For the adverse In Case 1
the proportions are such that only the straight portion of the secorid polar is
relevant.
In Case 2 only the straight
portion of
thefirst polar is relevant. It was not
until
Case 3 was reached that the idea was mentioned of using two polar curves which touch,xit in
fact
Casos i
and 2 use the same process.One further examples of sailing to windward is instructive as showing that the polar curve method can be applied to charges in conditions resulting from differential. current movements.
Case 4 is shown in FIg. 6 where M is the windward mark, moored in an adverse tidal stream, and to be reached in a wind steady both
in
strength
and direction. There is,however,
a strip of water
unaffected by
the tide
andobviously
the yacht should tack up in this until the mark canbe reached
in a single tack. ini 11g. 5 the polardiagram
for still water Is shown with origin 01 and thecurrent section the polar diagram is altered for two reasons.
F.. L the adverse current reduces the vîind relative to the
sea surtace, and the polar diagram relative to the sea is therefore
tu
be drawn br a reduced windspeed, as
shown by the line H2 A9 C2 with origin O.. NOW take a new origin 02 such that02
0represents the
current, and
the polar diagrdm relative to the markwill then be the . lue H2 A9 C2 with origin O:,
The problem is to sail from a given starting point, say
01,
to M in the shortest çosible iine. if Mis parallel to At,, 0a course such as 0 Q1 M is possible (the yachts
course throuqh th water on the lE-q M bein9 parallel to 01 A2). There is,
a faster course such as
01 T2 Q2 M, on Which the yacht is uled in the normai manner to Q, and then 'from to M wide of the
normal. windward courses The vector ovei' Lhe ground on this leg
will be
02
P and through the water1.
The proof of the assertion that. such a faster course exists will
nnt
he given here, but requires on .y the ssumptJon (alreadymentioned -as feature (iii) of those generally obtaining) that the curve A, P l-{, is tn'entiai to, and does not
start off at an angle ¿
with, -the straight line C2'A2. it is then posibìn to show that
the couxse to Q1
and thence direct 'Lo M 'is actua?ly slower than one which can be found caiLing th final 1-eq slightly free. Given theactual polar curves and the current velncity the problem could be
crnnp.Letely
solved.I
I
Sailing Down Wind
The expression "tacking to leeward" is
11 known, but the
practice of this manoeuvre is not as frequent as it cught to bc.
Let us first consider the case of a yacht with
a very high reachtn.qspeed. The ice yacht Is the supreme example, but the sama considerations can
arise w:ith a
catamaran or a sailing canoe, aridfor other light
displacementcraft.
The po1r curve of
performance may have a hollow in the duwrì winddirection as in
Fig. 7, and if this is the case it is obvious that In a steady wind any down wind leg of the course within the sector FCX should besailed
tacas parallel to OF and
,
and the proportion of time
spent on thase two directions adjusted to give the
required
resultante
Hovvor, the yachts to which this applies are rare.
But a
change ofwind direction may make
tacking to leeward proftahte inany yacht, even tf the polar curve is like that showi :n FIg. 1.
The argument is similar to that already applIed In Case 3.
Case
is illustrated in Fig. 8.
The leg to be sailEd Is
from M1 to M.,
with wind dlreçtlort initially straight down the page,But befoe It Is possible to reach M9 in this wind, a clockwise change
of direction takts place.
The polar curve is drawn with
as
origin to the scale which represents the locus of pints attainabte
within th
time before the wind
changesdirection.
From M, arid
for the naw wind direction a reversed polar is then drawn to such a
sca].e thti it just touches the first.
The point cf contact, P,
io
-to be sailed arc M1 P and
This example shows that a
change of wind direction may make "tacking to leeward" advantageous in certain circumstances, even when tho yacht'spolar may
notsuggest
this 9fl first inspection.( in Fig. shows that it is possible for a straight course
to be advantageous even
when the wind changes direction,but this
I
will be comparatively rare, as it occurs by chance only for certain relative positions of M1 and M2.
To
conclude this series of examples, Case
7 is onewhere
themost advantageous course is
not a straight one because of a change
in nrid treng th. Fhis can arise because thepolar
curve for the same yacht is usually of quito a different shape in a light wind to that in a strong wind. Fig. 10 shows the form the polar curves commonly take. The onefor light wind
is of a different shape, and in Fig. 11 theresult is shown when the log to he sailed is a broad
reach
and the wind falls from strong to light athalf time.
The pr.coss and thelettering is exactly the same as for Case 5, Fig. 8. The result, we 1 known to
yachtsmen, is
tht ji pays to git to leeward 'f the direct course while the wind is strona.Certerol Discuscion
lt cviy be
asked how
these polîr curves hivc been measured, and r must confess tho'l.they
have not. rheir shape is based on the eccum'jLtcd experience of yachtsmen, and to sorne extent ust1fied byth sucessfu.1 application of the
results to which
they 1ad. The purpose of thispaper is
to show that knowledge of tha completeI
out of the best way to get round a course in predicted wìnd and tide conditions. The complications are formidable even when the polar curves are available, but all know so far as yachtsmen is that afew generai
principles bring success - for example,that it is
better to go
quickly somewhero
than slowly on the direct course. This maxim is illustrated by several of the examples already given.I hope that i may focus
attention on the
need formeasurcrnents
of the polar curves of a number of representative yachts, arid for a more detailed study along the lines described, but using po]ars a.actually measured. Only in
this way
can it be ascertained with certaintyhow much wide of
the normal close-hauled course oneshould
sail in given conditions, or how far to leeward of the direct course one should sail on a broad reach when expecting the wind to die away.
The work which is taking place at Southampton University will no doubt provide comprehensive information of the kind required, arid a
study of the polir curves and their behaviour when used in ossihie cases of changes of wind direction and strength is the only way to reduce to a scientific
basis the
art of sailing in ivariable wind.
The same methods require extension to the cases where
the wind
direction varies, not with timo but with position, and the problem jsto
pick the most advantageous course through such variations.The complications are no more formidable thon
have
beendeì.t
with in other fields -for
example, studies of the art of gliding for maximum duration, orfor maximum
range in a head wind, Yacht;men as- 12
-a whole h-ave le-arnt the hard way a great deal which could have bean predicted ori the basis of such polar curves had they been available., and if ty can be measured, or computed from otherperformance measuremento, the nethods evolved by yachtsmen are capable of further refinonent.
i
i
Ad dendurn
Five years have elapsed since this paper was written aiìd there is
little to
add. However the development othe C class
Catamaran has enormously increased the need for fullunderstanding of "tacking ta leeward".
These cro lt have deve IOL:U the una rig o the most effective overall, but in iigìt or moderate winds the L:st. downwir. performance is on a course at approximately
450 ta
th& true wind direction. The polar curve i.s similar tn
rig. 7 but
the angle k0G is about 900 and the hollow between E and G is very marked.The result is Lhat coisiderations similar to those descrìhe in cases, 1, 2 and 3 apply also to dQwn wino sailing, and if the wind is fluctuating in direction abut a fixed average, then whenever
tfl
wind has shifted clockwise of the averagc the yacht should be on the port qybe, and marcver sriouid be sailing slightly higher than the noimal angle tor Leeward saiLing in a steady wind.The well known working rule tar windward sailinu; "Tack whenever the 'und heads you" how has a corresponding comparison
for leeward sailing which maj be expressed as "Gybe whenever
the wind comes more att". This rule must he
urierstad as Ieterriûg
to the true wind direction; and this must be carefullydistinguished
from that of the apparent wind. Generally speaking the difference in direction between the two i; with this te of yacht, nuch greater ir downwind than in upward sailing.
PJLAR CURVES FOR A J 2-4ETRE YACI{T
AtflENDU IT
S.U.Y.R. REFORT No. 20.
In order to aive a more precise idea of the form of the polar curves
for real
ychts, Figure 12 ha been prepreo. It shows estirnited polar
curves for a modern
12 mntre yacht silirg in calm wter in trua wind
strenqths at'7, 12 and 20 knots,
carrying genoa or spinoake.r as 3propriate.
The curves are in reasonable agreement
with observations ot 12 metre performance as Lve11 as with the
performance eected
rm tank tests 01 hulls ¿md wind tunnel tests of the seiLs.
However as both fufljze
ohservatiors arid predictions frocs model tests are subject to error these curves must
. .only
be
regarded asapproximate.
Furthermore iii open water the stronger winds would
ciuse a cons!ierble sea.
This would lead to a marked reduction in the speed made
qod to windward, and would also affect, to some extent, the speeds on
other headings.
Thofitr--
marked alongside the polar curves indicate the apparent
wind angle,
i.. the angle between the yocht's centreline and the direction of
the wind
elative
le.
the yacht.
This is the angle that would be indicated by a wiridvane on the yacht,
could he located at a positiun where the air flow was not affected
by the
rsdmity ot the sailc.
It is ir1t?restinq to see
much the forward speed ot th
yr;ht causes the apparent wind to draw ahead, especially it the lower wirìdspeeds.
Theof the yacht also affects the velocity of the
apparentwind, which is reduced
rDsome three knots when running in a 7 knot true wind.
This results iii a holl
ir
th' donwi.rd
part of the polar dia.ram which becomes less marked as the wind velocity
boccncs laroer relative to the yacht's speed.
The form of the polar cuve
ts sonewhat dependent on the
fact that the
12-metre is a yacht cf rather narrow beam, rather large
displacement/length ratio
ur her size arid hig
ballast, ratio.
Speeds up to about 7- knots
(V,4rL
1.1)
.in thus be achie'ed relatively easily, but thereafter the resistance rises increasingly
ridly and vory lar;e toxces are required to drive the hull above
about lO knots
1.5).
Hulls of lower dtsplacernnt/length ratio, more easily driven at high
speds, would Lravel laster in broad reaching conditions and
would show a hollow In
the runninQ part of Ihe polar curve at higher windspeeds than the
12-metre.
TheCataniaran referred
us earlier is an extreme exampie of this
behaviour.
For other displacement yachts the polar curVe5 would be of broadly similar
form
to those for .he 12-metre.
For smaller yachts having hull and
rig
characteristics silar to those of the 12.-metre, polar curves
giving the
approximate perto:ance might be obtained from those for the
12-metre by reducing
'J
and V. values iii proportion to the square root of the waterline
lengtho
WIND
FIG.1
POLAR CL!V FOR GIVEN
WIWO SPEED
WIND
FIG.2
sìj.
¡VII/LE
P
HT
SLACK
WATER
WIND DIRECTtOH AND
STRENGTH4 WiND FROM Mi
TO P
C-, I I-x
Io
I I I I W INWIND____
DIRECTIONWIND FROM
M1 TO P
C.)
/
WP4D
DIRECTOH