Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Computational fluid dynamics

(CFD) sail shape optimisation 13


B. Fallow
North Sails, Auckland, New Zealand

13.1 Introduction
This chapter will explain the process of sails design for a particular application.
When it is time to consider how to design a sail for a particular application, sail
designers consider the conditions that the boat will be sailing in, as well as the way the
boat will behave in those conditions.
This interface between the forces and moments the sails produce, and the way the
hull resists these forces and moments must be taken into account. In order to do this, a
velocity prediction program (VPP) is used to help establish the key parameters that we
need to optimise in order to create the desired flying shape of each sail.
The difference between design shapes and flying shapes for sails, is reviewed and
why understanding this is important to sail designers. Next we shall explore the pro-
cess sail designers need to go through in order to accurately predict what design shape
is required to get to a target flying shape. In order to do this there is a review of the
­aero-elastic effect that exists, and also a description of the computational fluid dynam-
ics (CFD) and finite element analysis (FEA) tools available to model this.

13.2 The right sail for the job


When most people see a boat sailing along, they have little idea of the battle being
waged to balance out the various forces and moments that are generated by the hull
and the sails. Sails provide drive force to a yacht through the mast, riggings and other
sheets that are attached to the hull. Drive is defined as the force in the direction the
yacht is travelling. However, except when sailing directly downwind, there will be a
significant side force produced as well, which the hull and keel must resist. This is be-
cause sails generally produce more lift (defined as a force 90° to the onset wind flow)
than drag (force in the direction of the onset wind flow).
Entire books and careers have been based on this topic (Claughton et al., 1998), but
in the very simplest terms, the yacht heels over until its righting moment matches the
heeling moment of the sails, and the boat speeds up until its drag through the water
matches the drive force of the sails. So designers think of a yacht as a force and mo-
ment balance machine.
Over the past 30 years, VPP have been developed to model this balance of forces
and moments for yachts sailing in steady state conditions (Kerwin and Newman, 1979).

Marine Applications of Advanced Fibre-Reinforced Composites. http://dx.doi.org/10.1016/B978-1-78242-250-1.00013-2


Copyright © 2016 Elsevier Ltd. All rights reserved.
296 Marine Applications of Advanced Fibre-Reinforced Composites

As time and u­ nderstanding have gone on, these have become more sophisticated and
more accurate. In the 1980s, a VPP would only balance the drive and side forces acting
on the yacht, as well as the roll moment. Today's most developed VPPs solve for all
three forces (drive, side and heave), as well as all three moments (roll, pitch and yaw)
(Claughton et al., 2012).
VPPs become particularly valid in sail shape optimisation where they can be used
to predict the preferred force and where it acts for a given set of sailing conditions and
hull design parameters.
If we assume that we could use a VPP to deliver an optimal set of sail forces for a
given condition, then we need to work out how to design a sail that, under load, will
yield those designed force and moment properties. This is the process of sail design,
and we shall discuss this in the remainder of the chapter.

13.3 Sail shape parameters


When looking at a sail, there are four key parameters that will define the surface of
the sail.
a. Planform. The planform is the overall outline of the sail plan. It is bounded by the mast
height and boom length, as well as the location of the headstay, both up the mast and where
it attaches to the hull. These are often described as I (fore triangle height), J (fore triangle
base), P (mainsail luff length) and E (mainsail foot length), as shown in Figure 13.1. Further
to this, the planform is defined by the profile of the Leech or aft edge of the mainsail (defined
by a series of girth measurements) and more recently the width of the head of the main is
a factor as well as so-called ‘Square-top’ mainsails have become fashionable (as depicted

P
I

E J

LP

Figure 13.1  Principal sailplan dimensions.


Computational fluid dynamics (CFD) sail shape optimisation297

in Figure 13.1). For the headsails the amount of overlap is important, and this is defined as
the distance perpendicular to the Luff or leading edge of the headsail. This measurement is
known as LP and is usually expressed as a percentage of the J measurement. This is referred
to as the overlap of a headsail.
Planform is often dictated by the rules and regulations a yacht is sailing to (if rac-
ing), and in general racing yachts tend to have larger sailplans than a yacht designed for
cruising.
b. Camber. Camber is the main power function of a sail. Camber is defined as the maximum
depth of a section of sail in the direction of the flow, expressed as a percentage of the local
chord length of the sail at that particular section. This value can vary widely depending on
the power requirement of a boat, and the general trend is boats with larger sail plans and/or
low righting moments will require sails with less camber than stiffer boats with small sail
plans. Also boats sailing in light wind strengths will require more camber to transfer enough
power to the hull.
Although there will be variances depending on the boat and wind conditions, sail
designers expect to see mainsail cambers vary from around 13% in the light air settings,
to as flat as 6% in an overpowered state. For headsails, the range is from as deep as 17%
camber in light airs to as flat as around 10%. One way to justify the fact that mainsails
are flatter than headsails is to think of the mainsail as a flap on a wing. Also it is worth
pointing out that it is possible to adjust the depth of the mainsail with a single sail as it
is attached to the mast by manipulation of the mast structure, which affects the depth of
the mainsail. Conversely, a headsail can usually only be adjusted by 2% or 3% of cam-
ber by sail trim manipulation, so we often see a range of headsails targeted for different
cambers throughout the range desired. We shall discuss the means of manipulating the
sail to achieve the desired camber later in the chapter when we explain how we use FEA
structural models in sail design.
c. Sectional shape. The sectional shape is an extension of the camber concept, whereby rather
than just defining the overall depth of a section of the sail, we are talking more about the
overall profile of the sail at various heights. Key parameters here are draft, front and back
percentage, all of which help describe the shape of the profile, Figure 13.2. For many years
we have been analysing these shapes with horizontal lines or trim stripes on the sails. Several
software programs collectively known as Sailscan programs exist to analyse these shapes
and a typical result is shown in Figure 13.3.
d. Twist. Twist is a parameter which describes how much the straight line chord between the
luff and leech of a sail changes with respect to the centreline of the boat as you go up the sail.
Like camber, twist is a power function as it can be viewed as a means of changing the angle

Fwd% Camber
Back%

50% 50%
50% 50%

Camber or draft position

Figure 13.2  Sail section shape.


298 Marine Applications of Advanced Fibre-Reinforced Composites

Advanced sail analysis software File name: M-2 7 Kts 2011 Pic 2.JPG

Ver.5.616, Copyright (North Sail Japan)

18.0 Camber 60.0 Draft 30.0 Twist


16.0 55.0 25.0
14.0
50.0 20.0
12.0
45.0 15.0
10.0
40.0 10.0
8.0
6.0 35.0 5.0

4.0 30.0 0.0


0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100
70.0 Entry 15.0 Exit 100.0 FORE_CAM
60.0 10.0
90.0
50.0 5.0
40.0 0.0 80.0
30.0 –5.0 70.0
20.0 –10.0
60.0
10.0 –15.0
0.0 –20.0 50.0
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100
SailType; Main Blue...Actual
Height Camber Draft Twist Entry Exit FORE_CAM BACK_CAM
25.0 6.3 49.6 3.8 11.8 –8.1 63.3 70.7
50.0 7.7 45.1 7.6 20.3 –6.1 69.8 71.9
75.0 7.1 41.8 11.5 29.4 0.6 75.2 68.7
87.5 5.3 43.5 12.9 28.9 4.1 80.3 70.2

Figure 13.3  Typical output from a Sailscan program.

of attack of a particular section. This effect can also be seen in Figure 13.1. Aside from this,
there are some key aerodynamic reasons for twisting sails. First is due to the fact that a yacht
sails through a wind field where the speed of the wind varies with altitude. As boats sail at
an angle to the true wind, the effect on the wind profile the boat actually experiences (called
apparent wind) is that the apparent wind speed and the apparent wind angle (AWA) both
change with altitude. This effect can add up to several degrees over the height of the mast.
Computational fluid dynamics (CFD) sail shape optimisation299

Additionally there is an aerodynamic upwash effect due to the sails generating lift, and the
fact that the sails will shed vortices at the tips of their span. This effect is also seen in airplane
wings. This effect can be quite large, resulting in around 10–15° of upwash over the span of
the sail. Therefore adding these effects together, it is common to see sails twist by between
12° and 25° over their span. Add on top of this any adjustment the sail trimmer may want to
increase or decrease the power function of the sail.

The four main flying shape parameters to describe the shape of a sail are now de-
fined. It is critical to understand that all of these effects interact with each other. For
example, if you change the twist of a sail, then both the camber distribution and the
sectional shape will be affected. The only absolute is the planform, as that is defined
by the physical form of the sail material.
Clearly in order to understand this complex interaction of the key shape parame-
ters, sailmakers need to effectively model the structure of sails in order to make reli-
able predictions of how a sail will behave in normal use, and hence be able to meet the
requirement of the yacht as specified by the VPP.

13.4 The aero-elastic effect in sails


In most applications where performance is desired, it is desirable to build sails as light
as possible. There is a balance to be reached, of course, in effectively capturing the en-
ergy of the wind and transferring it through the mast to the hull of the yacht. Also sails
must be able to tack from one side of the yacht to the other, and in most applications,
it is a requirement of the regulations that sails are thin and only made from a single
surface (and typically only three corners are allowed).
As such, the structure of a sail, when acted on by the pressure of the wind field
deflecting around it, will deform, and as such the shape of the sail will change as well.
Now because the sail has changed, the pressure field will also have been altered, lead-
ing to a modified sail shape. This is a classic case of aero-elastic coupling as shown
in Figure 13.4.

Input
shape

Calculate Aero-elastic coupling


aero loads Deformed shape

Trim / FEA
settings

Figure 13.4  Aero-elastic coupling as related to sails.


300 Marine Applications of Advanced Fibre-Reinforced Composites

In sail design terms, we describe the initial unloaded shape as the ‘Moulded Sail
Shape’. The deformed shape after the effects of the pressure field acting on it is known
as the ‘Flying Sail Shape’. In simple terms, sail designers look at the moulded shape,
whereas sail trimmers on board the yacht look at a flying shape.
The key job of the sail designer is to bridge the gap between Moulded and Flying
Sail Shapes. For years this was based on experience and intuition, but over the last
10 years we have seen the rapid adoption of FEA programs to perform the stress
analysis, coupled with CFD programs performing the aerodynamic pressure field cal-
culations. The first package to do this effectively was FLOW/Membrain (Anon., 2010,
2014), and this has been under constant development for the past 25 years. However,
within the past 10 years, increasing computer speeds has meant that the packages use
has become a daily activity.

13.5 Predicting flying shapes


The sail designer's main job is to be able to effectively predict how a sail will behave
on the yacht. In order to do this with a degree of confidence, an accurate model of
the entire yacht above the deck is required. This modelling includes the sails, mast,
rigging, and all the sails and mast controls lines and fixings. As well as modelling
the mast and sails, structurally, we must also model the aerodynamics of the mast/
sail package. Finally, we need to simulate how sailors on board the yacht will trim the
mast and sails.
Therefore, to effectively model the environment of a typical sailing yacht, we must
accurately model the following four key inputs.
a. The unloaded geometry. The first input required comes from the sail designer, and this is the
shape of the unloaded sails. This was previously described as the Moulded Sail Shape. The
basic mast geometry also needs to be defined, with particular care to make sure that attach-
ment points for different rigging elements are correctly placed.
b. Structural properties of the mast and sails. Here we need to state the materials, along with
the key structural properties (EA, Poisson's ratio, etc.), as well as the orientation of the
material in sails. For traditional panelled sails this is relatively simple to define. However,
for more modern String sails, or the most recent 3Di Composite structures, as described in
Chapter 14, this can be quite complicated. The main point here is that all structural compo-
nents of the sail are incorporated, including battens and other fittings. For the rig, the key
bending and compressive properties (EI and EA) must be defined for the mast tube as well
as for all rigging components.
c. The pressure field acting on the sails and mast must be defined. Note that because sails are
thin by nature, then all we need to evaluate is the pressure difference from one side of the sail
to the other. The actual pressures on the surface are not required. This is where we use our
CFD program. For cases where a yacht is being modelled that is sailing at a narrow AWA,
say less than 40°, it is acceptable to use vortex lattice methods to calculate the pressure dif-
ference from one side of the sail to the other (Richter et al., 2003). For sails that are modelled
at AWA's greater than 40°, we need to use more sophisticated CFD programs such as RANS
(Reynolds Averaged Navier Stokes) codes, which can model separated flows (Ferziger and
Peric, 2001, p. 292).
Computational fluid dynamics (CFD) sail shape optimisation301

d. The boundary conditions. Here we need to define all the real world adjustments that we
would do to the mast and sails. This starts with what is known as a dock tune. This represents
the amount that the shrouds and stays supporting the mast are pre-tensioned such that the
mast tube will remain stable under load. This information is best supplied by a mast man-
ufacturer, as they are the experts in this particular field. For the sails we need adjust all the
controls that a crew on board the yacht would do. For a headsail this would include the Jib
Sheet, Halyard, tack or Cunningham and car position. For the mainsail we would manipulate
the Sheet, Cunningham, Halyard and traveller. For the mast, aside from the components that
were pre-set for the dock tune, there are additional active controls (known as running rig-
ging). The controls can have a dramatic effect on the bend of the mast and the tension on the
headstay. Both these effects can have a dramatic effect on the Flying Sail Shape, and must be
modelled accurately. These controls include backstay, checkstays, mast butt, deck chocks,
headstay length and running backstay, as described in Chapter 12 (rigging).
Once these four categories of data have been entered we can run the model through an FEA
program. The FEA program will yield two main outputs.
a. Load data. Once converged, the FEA program will yield all the loads in the rigging, mast and
sails. The loads in the standing rigging and running rigging elements will be expressed in
Newtons, while the mast tube will be able to additionally yield bending moment and torques.
These loads can be as useful to mast designers as well as sail designers. The sails load at the
attachment points (sheet, halyard, mast) can be derived, as well as the load in the individual
components that make up the body of the sail (yarns, filaments, films). This can be viewed
in one of two ways: first as ‘Yarn Strain’. Yarn strain is the load in the direction along each
load bearing yarn. Analysing yarn strain output allows us to ensure that no single component
of the sail is experiencing too much strain (typically 0.2% strain for Carbon sails). The other
type of strain output that we can use is called ‘Element Strain’. Here the net effect of all the
different layers that make up the sail at a particular point are analysed, and the total strain
and direction of deformation are reported. The simplest way of describing this output is to
imagine how the surface of the sail will deform at a given location, rather than looking at
what is happening to the individual components that make up the laminate or composite at
that point.
b. Flying shape. The converged FEA run will yield the deformed shape of the mast and rigging
elements, as well as the deformed or Flying Sail Shape. This shape can be expressed either
as a 3D surface file, or in terms that sailors commonly use, such as a Sailscan output (see
Figure 13.3).

Note that once we have the Flying Sail Shape, we are now able to close the
a­ ero-elastic loop, by sending the deformed sail shape back to the CFD program, and
then running the FEA program again. Experience has shown that this iterative ap-
proach as shown in Figure 13.5, is very stable, and convergence is reached quickly.

13.6 Closing the design cycle


In the previous section, we have seen how we can model the difference between the
moulded sail shape and the Flying Sail Shape. In order to close the design cycle, we
need to know what Flying shape is desirable for a given sailing condition. To achieve
this we need to return to the VPP.
302 Marine Applications of Advanced Fibre-Reinforced Composites

Major strain
Unloaded 0.005000

sail/rig
0.004286
design
0.003571

0.002857

0.002143

Mast / sail
structure 0.001429

0.000714

0.000000

Aero forces Aero-elastic coupling


(CFD) Flying shape

Trim / FEA
settings

Loads

Figure 13.5  Inputs and outputs of coupled FEA/CFD mast/sail simulation.

It was stated earlier that early VPP's had only three degrees of freedom, with very
limited inputs for sails. Generally sails were represented as simple geometry areas
with a Reef or Flat function to modify standard Force coefficient look up tables. More
recently, VPP's have been developed that will solve the forces and moments in more
axes, and they have allowed more sophisticated models of the sails to be incorporated
in the solver as a variable. An example is the NorthVpp, in which variables for sails
such as Camber, Twist and Sheeting angle, mean that it is possible to input Flying
shapes, then let the VPP decide on the Optimal trim angle. In order to generate the
Forces and Moments that the VPP will use for the sails, a Vortex Lattice CFD program
such as FLOW is used. Figure 13.6 shows a typical screen capture from flow.
A batch of runs is set up for a variety of heel angles and AWAs for each set of sail
trim settings. This creates a multi-dimensional space of sail forces and sail moments
that the VPP solver can balance against the hull forces. The result is the optimal Flying
Sail Shape for a given set of conditions.

13.7 Conclusions
We set out to show how CFD can be used as a component to designing better sails.
Along the way we have seen that in order to do this effectively, we must first consider
the difference between a ‘Moulded Sail Shape’ and a ‘Flying Sail Shape’. Coupled
Computational fluid dynamics (CFD) sail shape optimisation303

Figure 13.6  FLOW Vortex Lattice software run set up for VPP case, with sail trim variables
active.

FEA/CFD programs such as FLOW/Membrain are essential to do this on a consistent


basis, provided you are very thorough in your modelling of both the sails and the
mast. Once our flying shapes have been defined, then we can use a VPP to predict the
performance of the sail, and ultimately, the optimal sail for a given boat, in a given
sailing condition.

References
Anon., 15 October 2010. North Sails Design Video. https://www.youtube.com/
watch?feature=player_embedded&v=ErP1x8XLtUo.
Anon., 2014. Sail Design. http://www.uk.northsails.com/TECHNOLOGY/SailDesign/
tabid/6912/language/en-US/Default.aspx.
Claughton, A.R., Wellicome, J.F., Shenoi, R.A., 1998. Sailing Yacht Design: Theory. Longman,
Harlow, ISBN: 0-582-36856-1.
Claughton, A.R., Pemberton, R.J., Prince, M.P., 2012. Hull-Sailplan balance, “lead” for the
21st century. In: 22nd International HISWA Symposium on Yacht Design and Yacht
Construction, Amsterdam, 17–18 November 2012.
Ferziger, J.H., Peric, M., 2001. Computational Methods for Fluid Dynamics. Springer, Berlin/
Heidelberg, ISBN: 978-3-540-42074-3. http://dx.doi.org/10.1007/978-3-642-56026-2.
Kerwin, J.E., Newman, J.N., 1979. A summary of the H. Irving Pratt Ocean Race handicapping
project. In: 4th Chesapeake Sailing Yacht Symposium, 1979.
Richter, H.J., Horrigan, K.C., Braun, J.B., 2003. Computational fluid dynamics for downwind
sails. In: 16th Chesapeake Sailing Yacht Symposium, March 2003.

You might also like