High Performance Yacht Design Conference

Aukland, 4-6 December 2002

WORKING LOAD TO BREAK LOAD:

SAFETY FACTORS IN COMPOSITE YACHT STRUCTURES

Giovanni Belgrano MSc1, [email protected]

Luke McEwen MA2, [email protected]

Abstract. The loads imposed on yacht structures fall broadly into two categories: the distributed forces imposed by the action of the
wind and waves on the shell of the yacht, and the concentrated loads imposed by the rig and keel to their attachment points on the
structure. This paper examines the nature of the latter set of loads and offers a methodology for the structural design based on those
loadings.
The loads imposed on a rig attachment point vary continuously while the yacht is sailing. Designers frequently quote "working load",
"safe working load", "maximum load" or "break load" for a rigging attachment, but the relationship of this value to the varying load is
not always clear. A set of nomenclature is presented to describe clearly the different load states from the "steady-state" value, through
the "peak, dynamic" value to the eventual break load of the fitting and of the composite structure.
Having defined the loads, the structure must be designed to carry them with sufficient stiffness, strength and stability. Inherent in
structural engineering is the need for safety factors to account for variations in load, material strength, geometry tolerances and other
uncertainties. A rational approach to the inclusion of safety factors to account for these effects is presented. This approach allows the
partial safety factors to be modified to suit the choice of material, the nature of the load and the structure and the method of analysis.
Where more than one load acts on an area of the structure, combined load cases must be developed that model realistically the worst
case scenario. In particular if the loading is quasi-static, the total loads on the structure must be in equilibrium. This is particularly
important for Finite Element Analysis since an unbalanced load case can lead to excessive reactions at the points of restraint of the
Finite Element Model. A method is presented for the development of a balanced load case for upwind sailing which allows
significant insight into the behaviour of a yacht structure under "real" sailing conditions. The keystone of this approach is a method
for constraining the model in a statically-determinate manner, to avoid adding unrealistic stiffness to the model.
Finally, once the structure has been built, it is sound practice to proof test it to give confidence in its reliability. The value of load for
proof testing is a difficult choice but is made more straightforward by the rational approach to load definition presented in the paper.

1. INTRODUCTION “working load”, “safe working load”, “maximum load” or

The loads imposed on yacht structures fall broadly into “break load” for rigging attachments, but the relationship
two categories: the distributed forces imposed by the of these values to the varying load is not always clear. A
action of the wind and waves on the shell of the yacht, set of nomenclature is needed to describe clearly the
and the concentrated loads imposed by the rig and keel to different load states from the maximum “steady-state”
their attachment points on the structure. The value, through the “peak, dynamic” value to the eventual
concentrated loads are relatively straightforward to break load of the fitting and of the composite structure.
measure or calculate, allowing engineering calculation of The system shown in Figure 1 has been used successfully
the structure required to carry them. The distributed for two decades for the design of composite yacht
loads are much more difficult to define, so are usually structures. It can be summarized as follows:
dealt with by designing the structure to a classification Table 1. Linked to figure 1
society rule. This paper examines the nature of the
concentrated loads and offers a set of nomenclature to W1 Maximum steady-state load (flat
describe how the loads vary as the yacht sails along. water)
W2 Peak dynamic load (due to waves,
This paper covers the reasons for using safety factors in
gusts of wind, manoeuvres, sudden
structural design. Safety factors used in marine easing of sheets etc.)
engineering, particularly by classification societies, are
often hidden in the formulae used. Furthermore, most LIMIT Elastic limit of composite structure,
classification societies avoid defining the loads acting on Break load of rigging rod, fitting
the rigging. This paper suggests a clearer methodology etc
for incorporating safety factors into the loads and ULTIMATE Break load of composite structure
material properties used for design, both by traditional
methods and by Finite Element Analysis (FEA), and for
the subsequent testing of the yacht structure to check its It should be emphasised that the W1 and W2 loads are
strength. the real or anticipated loads that will be applied to the
structure, in other words they are the inputs to the load
calculation. The LIMIT and ULTIMATE loads are the
2. THE NATURE OF RIG LOADS loads that the structure is designed to withstand, so are
The loads imposed on a rig fitting vary continuously the inputs to the structural design, and are calculated from
while the yacht is sailing. Designers frequently quote the W2 loads.

1 Marine Manager, SP Systems

2 Senior Engineer, SP Technologies
 tensile or compressive) stresses in particular will be
ULTIMATE sensitive to small changes in geometry.
LIMIT d) Accuracy of the analytical method: a well-executed
Force

Finite Element Analysis (FEA) may allow the calculation

W2
of deflections to within a few percent of the true value,
W1 but stresses will be less accurate. Traditional calculation
Load
methods will usually give even larger errors, particularly
Time for geometries that do not readily simplify to cases with
an explicit analytical solution. Stress concentrations due
Figure 1. to features not considered in the analysis (e.g. holes for
The forces on a rig attachment vary continuously as a fasteners) are probably the most common cause of
yacht sails along. With a load cell in suitable weather it inaccuracies in the calculation of stress.
is reasonably straightforward to establish the W1 e) Other effects that might not be considered explicitly in
(maximum steady-state) load. If that load cell is the analysis could include fatigue, creep, environmental
connected to a data logger, after many hours of sailing the effects (ageing), pre-stress due to manufacturing methods,
W2 (maximum dynamic) load can be established. damage in service, and so on. While these effects should
The W1 load is frequently reached but the real load be taken into account if they are going to have a
varies continuously around it. However it is a convenient significant effect on the structure, often it is considered
load to define since it is easily measured. The right value sufficient just to use a factor of safety to cover them.
for the W2 load is more difficult to establish; even if the Clearly the choice of safety factor is critical to ensure that
load is measured for several years, there is no guarantee the structure is stiff, strong and stable enough in service
that the highest measured load would not be exceeded throughout its design life, without being “over-
during the remaining lifetime of the yacht. engineered” to the point where its weight or cost
The most rigorous solution to this problem, pioneered by (performance or financial) becomes detrimental. In
the aerospace industry, is to use statistical extrapolation assessing this balance, the engineer is making a judgment
to determine a W2 load that is sufficiently unlikely to be on the consequences of failure. In the aerospace industry
exceeded in the lifetime of the yacht (or aircraft)[1]. The this process is taken to its ultimate conclusion: safety
more data is available, the closer this theoretical W2 load factors are chosen to achieve an “acceptable” number of
will be to the highest recorded peak load, for a given fatalities per passenger mile (hopefully a very small
level of confidence. In the marine industry, years of load number)[1].
measurement would be prohibitively expensive and time- In the yachting industry structural failures are more
consuming, so fewer measurements are taken and the W2 common, which is considered acceptable because they
load is chosen to be significantly higher than the highest are less likely to cause death or injury. In particular,
recorded peak load. rigging failures at a perhaps surprisingly high rate are
accepted because the performance cost of making rigs
“unbreakable” is too high and collapse of a mast is
3. SAFETY FACTORS: WHAT ARE THEY FOR? unlikely to result in fatalities. Similarly, the
Factors of Safety have also been called Factors of consequences of failure depend on where the boat is
Ignorance. Their purpose [1] is to account for all the sailing: collapse of the composite structure is more life-
reasons that might make a structure fail if it was threatening in the Southern Ocean than in the Hauraki
theoretically only just strong, stiff and stable enough to Gulf. Thus margins of safety can be pared down much
withstand the expected loads. These reasons might more in an America’s Cup yacht than in an Open 60 for
include: the Vendée Globe.
a) Uncertainty in the load data: there is a small but real
possibility that the load might exceed the chosen W2 load 4. HOW ARE FACTORS OF SAFETY APPLIED?
during the lifetime of the yacht.
There are three common ways to incorporate safety
b) Material variation: test values will always show some factors in engineering analysis[2], illustrated in Figure 2.
spread, but statistical methods can be applied to material
test data to calculate the minimum strength of a material
with a certain level of confidence. Composite materials 4.1 Global Safety Factor: the simplest approach. One
in particular will show a wide spread of strengths, due to factor of safety is applied that accounts for every reason
variations in void content, fibre volume fraction and resin why the structure might be less stiff or strong in practice
mix ratio incorporated into the material during the than in theory. Typically, engineers tend to use a factor
manufacture of the component. between 2 and 6, although factors of safety of 18 or
c) Geometric tolerances: a certain geometry will be more[5] have been used for very safety-critical
assumed for analysis, but building methods will mean that applications where there was considerable uncertainty
the real structure will be different to some extent. about loads or material strengths. The factor of safety
Analyses of buckling and of Brazier (through-thickness can be applied either to the loads (the “load factor”
method) or to the material strengths (the “permissible
stress” method). Whilst being extremely straightforward The loads and material properties used in the analysis
to use, these methods take no account of differences in would be “characteristic” values, i.e. chosen statistically
analysis methods, material variability, load uncertainty or to encompass all but the worst 5% or so of likely
consequences of failure. values[2].
This approach is more precise than the “global safety
factor” method in that it allows each influence on the
4.2 Limit State Design: the civil engineering industry
analysis to be considered separately. However it is
often uses several “partial safety factors”, each of which
relatively complex to apply in practice because each of
accounts for a particular effect. Some partial safety
the safety factors have to be applied to every calculation.
factors are applied to the load and others to the material
Note that the total safety factor on strength at the
strengths. The structural design must meet two “limit
Ultimate Limit State is 1.6 x 2.0 x 1.5 = 4.8, similar to
states”: the Ultimate Limit State and the Serviceability
the global safety factor that might be used in the global
Limit State[2].
safety factor method.
a) The Ultimate Limit State requires that the structure
must withstand the highest applied load without
collapsing catastrophically. This includes material 4.3 Simplified Limit State Design is an approach
failure, buckling or overturning. The partial safety pioneered by the aerospace industry and now used in the
factors for this limit state are relatively high. marine industry. It is less complex to apply than the full
b) The Serviceability Limit State requires that the Limit State Design method, because all of the safety
structure must not suffer excessive deflection, cracking, factors except those to cover the material variability are
applied to the loads. This is done at the start of the
fatigue, vibration, fire damage or other degradation under
its normal working conditions. The partial safety factors project. Thereafter, the safety factors need not be
consciously considered again.
for this limit state are lower.
For example, a bridge might be designed with safety a) The expected loads on the structure are expressed as
W1 and W2 loads (see Figure 1). The W1 and W2 loads
factors applied as follows for the Ultimate Limit State:
are not used for strength or stability analysis but are used
i) Partial safety factor of 1.4 on the “dead load”, i.e. the to check the structure for adequate stiffness.
self-weight of the structure and any snow or ice on it.
This accounts for uncertainty in the load, as the bridge Two (hopefully hypothetical) load states are then defined
for strength analysis, the LIMIT load and the
might end up weighing more than anticipated.
ULTIMATE load. At the LIMIT load there should be no
ii) Partial safety factor of 1.6 on the “live load”, i.e. the degradation of the structure, in other words it should
weight of the cars, lorries and people on the bridge. This continue to perform as designed. Beyond the LIMIT
accounts for uncertainty in the load and for dynamic load, the structure is allowed to yield, buckle, crack, etc
effects i.e. accelerations due to bumps on the road provided that it does not fail catastrophically until the
surface, walking loads etc. Because of differences in applied load reaches the ULTIMATE load [1,2].
dynamic accelerations, this factor might be less for a train
The LIMIT load is higher than the W2 load (the highest
bridge than for a pedestrian bridge (especially in light of
the problems with the Millennium Bridge in London). If load likely to be seen in service) by some factor, which
the live load tended to lessen the likelihood of failure accounts for all the issues raised above except the
material variability. There is a further factor of safety
(e.g. in the case of a stone bridge where the load might
stabilise the structure) a factor of 0.0 would be used. between the LIMIT and ULTIMATE load states. The
purpose of this factor is to ensure that, while LIMIT
iii) Partial safety factor of 2.0 on the material strengths. failures are rare, ULTIMATE failures should occur much
This accounts for ageing, fatigue, environmental effects, less often. This factor is normally constant for a given
strain-rate dependence of properties, pre-stress and structure, although it might be higher for an ocean-going
damage in service. The factor might be increased for yacht than an inshore racing boat.
materials with poor fatigue performance, poor UV
b) The material properties used for analysis are “design
resistance or for brittle materials.
allowable” values, which take into account the statistical
iv) Partial safety factor of 1.5 on analysis of strength and variations due to processing techniques, environmental
stability. This accounts for geometric tolerances and effects etc. Thus these properties should be reliably
inaccuracy of the analysis method. The factor could achievable in real structures built in a manufacturing
perhaps be reduced if FEA was used instead of traditional environment (as opposed to a testing laboratory) and
calculations. maintained over the expected working life of the
For the Serviceability Limit State, the corresponding structure. The above methods are compared in Figure 2
factors might be below. For the rest of this paper we will assume that the
Simplified Limit State Design approach is being used for
i) 1.0 on the dead and live loads the design of the yacht structure
ii) 1.0 on the material properties
iii) 1.0 on analysis of deflections and 1.5 on analysis of
cracking of facings etc
 fitting or rigging rod can then be used as the LIMIT load
ANALYSIS ULTIMATE for design of the composite structure, since the rigging
GLOBAL
PARTIAL FACTOR provides a “fuse” which will break at a reasonably certain
FACTOR
SAFETY load.
FACTOR
MATERIAL LIMIT c) In the case of a sheet or halyard, where the hardware
PARTIAL FACTOR
and ropes might be significantly oversized for stretch or
FACTOR
LOAD handling considerations, their break load would be an
SAFE PARTIAL W2 FACTOR excessive design load for the composite structure. In this
WORKING FACTORS
LIVE LOAD & situation the LIMIT load used for the composite structure
LOAD
DEAD LOAD should be just some factor above the W2 load, to account
for geometric tolerances and inaccuracies in the analysis
METHOD

SIMPLIFIED
W1 LOAD

DESIGN
FACTOR
GLOBAL
SAFETY

STATE
LIMIT
method or assumptions. For an aluminium airframe this

DESIGN
STATE
LIMIT
factor might be as low as 1.0[1]. Such a low safety factor
is justifiable only if the analysis methods are known from
test results to be accurate and conservative.
Figure 2: A comparison of methods for incorporating
safety factors in engineering analysis For yacht design, the time spent measuring loads and
designing the structure accurately enough to use such a
small safety factor is usually not worth the weight saving,
5. SIMPLIFIED LIMIT STATE DESIGN IN so the factor is usually much higher, perhaps 1.5 or more.
YACHT STRUCTURES d) The load on the composite structure cannot usually be
The steps required to apply the Simplified Limit State higher than the LIMIT load, since the rigging will break
Design approach to yacht structural analysis are at this load. For some structures therefore, it is sufficient
described below. The basic steps are: to make the composite strong enough not to break at
LIMIT load. In practice however, if the fitting does break
a) define all the loads that will act on the yacht in a Load (for some unforeseen reason), one does not want to have
Table to replace the hull structure in case it has been degraded
b) define the material properties to be used for analysis as in any way, even if it did not actually break apart. Thus it
a set of Design Allowables is usually wise to design for no degradation (e.g. resin
microcracking of composites, yield of metals, or other
c) analyse the structure non-catastrophic failure) at LIMIT load. This is
particularly important for composites as micro-structural
damage is so difficult to detect.
5.1 Definition of loads
e) An occasional local yield or microcracking failure is
At the start of the design process, a load table is drawn up
more acceptable than a catastrophic failure. To ensure
which gives the W1, W2, LIMIT and ULTIMATE values
that the statistical likelihood of a catastrophic failure is
for each loaded fitting (see Figure 1).
even lower than that of a LIMIT failure, some further
a) The starting point is the W1 (static working) load, factor of safety is required. The simplest way to do this is
which can usually be measured, estimated from to design the structure not to fail catastrophically at a
experience or calculated from first principles. The W1 hypothetical ULTIMATE load, which is greater than the
load is multiplied by a factor to get to the W2 (peak, LIMIT load by some factor. In aerospace, this factor is
dynamic) load. This factor varies according to the fitting typically around 1.5[1]. In an inshore raceboat, where the
under consideration. For instance it might be around 1.6 consequences of structural failure are less devastating
for a backstay chainplate, but only 1.2 for a forestay than they would be in an airliner, the factor might be
chainplate, since slamming into waves tends to tighten the reduced somewhat. On a blue-water cruiser, it might be
backstay but slacken the forestay, and because sudden significantly higher. Note that the ULTIMATE load
easing of the mainsheet will momentarily increase the cannot theoretically be reached because the rigging
load on the backstay. Given sufficient time and money, should break first, but it is a convenient tool for design
the W2 load could be calculated by measurement and purposes.
statistical extrapolation, but for most yacht projects the
only practical method is to factor up from W1 based on
experience. 5.2 Material design allowables
b) Whilst the W2 load is theoretically the highest load Test results will give a spread of values that can be
that the structure will see, it is prudent to specify rigging assumed to follow some statistical distribution (Normal[2]
and fittings that are somewhat stronger than this, to or Gaussian[1]) – see Figure 3. With sufficient test data,
account for fatigue, ageing, and uncertainty in the load statistical methods allow material property values to be
data. Hardware suppliers (e.g. Harken) tend to use a chosen which it can be assumed that nearly all future
factor of 2.0 or more above the steady-state (W1) load[6]. samples will exceed (say 90% or 99%) with a reasonable
Rig designers typically specify rigging that is at least 2.5 level of confidence (say 95%). In general, at least 5 test
times stronger than the W1 load[7]. The break load of the samples are required to give a reasonable level of
confidence in the results[3]. Of course, even if 99% of the on. Thus the analysis will show that if the structure is
material in a real structure is stronger than assumed, 1% strong or stiff enough to satisfy the chosen safety factor,
will be weaker. However, even if a small percentage of it will in fact usually be slightly stronger or stiffer still,
the material is slightly weaker than the assumed strength, and this extra is called the Reserve Factor (R.F.) or
the chances of this causing catastrophic failure are small, Margin of Safety (M.o.S.)[4]. These are defined as:
particularly where there are several (redundant) load Reserve Factor = (actual strength / required strength)
paths and if it can be considered that the ULTIMATE
safety factor includes a small margin to cover under- Margin of Safety* = (actual strength / required strength)
strength material[1]. – 1.0
*M.o.S. is usually expressed as a percentage
45 Thus a structure with a Reserve Factor of 1.05 could be
Test data
40 said to have a Margin of Safety of 5%.
No. of occurrences, n

35 Normal Because all the required safety factors have been

Distribution incorporated into the calculations, the designer should
30
25
always be aiming for a R.F. of just over 1.00 or a M.o.S.
of just over 0%. The tendency to design for higher
20 margins than this should be resisted; if the designer feels
15 more comfortable with a higher reserve factor, the safety
10 factors built into the loads were probably too small.
5 Any given structure will have at least two Reserve
0
Factors: the Reserve Factor over micro cracking or resin
shear at LIMIT load and the reserve factor over
Strength catastrophic failure at ULTIMATE load. If there is a
stiffness requirement at W1 or W2 load, there will be
another Reserve Factor over this criterion. Likewise if
Figure 3: Illustration of a strength-frequency diagram the given piece of structure is subjected to more than one
showing the statistical spread of strength values of a load case (for instance a keel structure subjected to
hypothetical composite material heeling and grounding forces) there will be LIMIT and
ULTIMATE reserve factors for each load case. All the
The strength of composite materials is more difficult to reserve factors must be greater than unity (see Figure 4).
predict than the strength of metals, because the material
itself is made as part of the component manufacturing Since the factor between LIMIT load and ULTIMATE
process. It is clearly important that the tested material is load for a given structure is usually kept constant, if the
made in a similar environment to the final component, LIMIT strength of the material (i.e. the yield strength for
ideally by the same people. Then the design allowable metals or the micro cracking or resin failure strength for
material properties derived from the test data should be composites) is low compared to the ULTIMATE
reliably achievable in the boat yard. strength, the LIMIT reserve factor will be the critical one.
Thus it is the material properties that determine whether
the structure is LIMIT or ULTIMATE critical. Knowing
5.3 Analysis the ratio between the LIMIT and ULTIMATE strengths
for each material saves doing both calculations, since the
Having defined the loads and material strengths, the critical case can be anticipated.
required scantlings of the structure can be calculated.
Every likely mode of failure should be considered, 3.00
including: 2.55

a) Excessive deflection at W1 or W2 load 2.15

Reserve Factor

b) Resin micro cracking or other non-catastrophic failure 2.00

1.50
at LIMIT load
1.04 0.95
c) Fibre failure, bearing failure, interlaminar shear failure, 1.00
through-thickness tensile failure, buckling, shear
crimping or skin wrinkling at ULTIMATE load.
0.00
GROUNDING

GROUNDING
HEELING

ULTIMATE
DEFLECTION

HEELING

ULTIMATE

6. RESERVE FACTORS AND MARGINS OF

LIMIT
HEELING
@ 30°

LIMIT

SAFETY
A Safety Factor is a number that is chosen by the
designer before the structure is designed or analysed. In
practice, structural materials come in discrete sizes: there Figure 4: Illustration of Reserve Factors and Margins of
are standard ply thicknesses, sizes of extrusions and so Safety for a hypothetical keel grillage structure consisting
of longitudinal and transverse beams. The deflection be restrained by putting it into a virtual sea and letting it
under 30° heeling load is critical for the transverse sail along until it reaches a steady state; in any case such
members and close to optimum. The ultimate strength in a model would take a long time to converge on a stable
grounding is critical for the longitudinal members, and solution. In essence, what is required is a mathematical
insufficient. way to represent the force of the water on the hull, such
that the buoyancy generated exactly balances the
“weight” of the model, the drag exactly balances the
7. COMBINED LOADS AND FINITE ELEMENT driving force from the rig, and the lift of the foils exactly
ANALYSIS balances the side-force.
One way to achieve this is to represent the water by a
7.1 Combined load cases number of spring elements connected between the yacht
model and the ground. As the boat is pulled forward and
With loads clearly defined in terms of W1, W2, LIMIT sideways by the rig forces and downwards by its own
and ULTIMATE, it becomes reasonably straightforward weight, the springs will stretch to react against the
to specify combined load cases. For instance, a shroud movement, and if there are enough springs, the net effect
chainplate with the D1 and V1 shrouds attached to it must will be something like the distributed forces due to
be able to withstand the break load of either rod. When buoyancy and drag. However the springs also add
one rod breaks, there will also be some load on the other stiffness to the boat, so as the forestay and backstay tend
rod, but probably not its break load. It would be to bend the hull, the springs will resist that bending and
reasonable to assume W2 load on the D1 shroud when the give the impression that the boat is stiffer than she really
V1 breaks, so the combined LIMIT load case would be is. To minimise this effect the springs must be made very
LIMIT V1 shroud load (i.e. the break load of the rod) “soft”, but the movement of the model in sink and pitch is
plus W2 load on the D1 shroud. This LIMIT case can then very large under the imposed forces.
then be factored up as usual to get the combined
ULTIMATE load on the chainplate. A better way is to restrain the boat with just enough
restraints to take out the six rigid-body degrees of
freedom (translation and rotation in each of the X, Y and
7.2 Load cases for FEA Z directions). If only enough points on the model are
restrained to remove the six degrees of freedom, no
This approach can be extended to specify load cases for stiffness will be added to the model.
Finite Element (FE) models. A useful system is to run
one “realistic” loadcase which includes all the W1 loads However, unless the applied forces are perfectly
acting on the structure under some steady-state sailing balanced, there will be some non-zero reaction forces at
situation, perhaps sailing upwind, as this is usually the these restraints, which could lead to unrealistic local
case with the greatest global bending moment on the stresses.
boat. In addition to this W1 loadcase, several LIMIT A successful solution to this problem is to restrain the
loadcases can be run, with LIMIT load applied to one model using the rigging (see Figure 5). It will be
fitting or rig attachment and W1 loads applied to assumed that the FE model consists of the hull and
everything else. That way, the stresses due to the LIMIT appendages, with the influence of the rig represented by
load are superimposed on the basic stresses due to the forces applied at the rigging attachments. Five of the six
global bending of the boat, the rig pretension and so on. degrees of freedom can be eliminated by restraining:
These two types of load case are explored in more detail a) The two V1 chainplates in the direction of the V1
below. shrouds
b) The forestay chainplate in the direction of the forestay
7.3 W1 Equilibrium Load Case c) The keel and rudder centres of lift in the direction of
Because all the W1 loads on the boat are maximum the lift vectors
“steady-state” loads, they should all balance out so that This leaves the boat unrestrained longitudinally. In
the model is in equilibrium, i.e. not accelerating in any reality the boat accelerates until its drag balances the net
direction. Thus the sideforce on the sails should balance driving force from the rig. The drag force when the boat
the lift from the keel, the mast compression should is at full speed can usually be assumed to act evenly on
balance the tension in the shrouds and the sheets, and so the wetted surface of the boat. Since the keel top is
on. This load case could therefore be called an approximately at the centre of the wetted surface for most
equilibrium loadcase. boats, and the drag force is small compared to the other
forces on the keel, restraining the keel top longitudinally
is a simple way to eliminate the remaining degree of
7.4 Restraints freedom. The stiffness of the keel structure means that
Finite Element models need to be restrained in space the small reaction force at this restraint causes only small
even if all the forces acting on them are in equilibrium. additional stresses on the model.
There is as yet no FEA code that allows a yacht model to
 Fortunately for the FE analyst, the pressure distribution
on a yacht hull in flat water is approximately hydrostatic;
FOREST
Y the local variations due to dynamic pressure head make
up a relatively small proportion of the net force.
V1 LWD Thus for the purposes of the FE model, the water can
usually be represented by a hydrostatically varying
V1 WWD pressure field. The water plane must be inclined to a
suitable angle of heel and trim to balance the assumed rig
forces and at sufficient sinkage so that the buoyancy
DRAG balances the weight of the yacht.
Note that the water pressure distribution assumed can
give a net force that balances the applied rig loads, but is
not locally an accurate representation of the water
pressure. In particular, the yacht’s own wave system will
KEEL LIFT reduce the pressure amidships and increase it towards the
bow and stern, tending to increase the global bending of
the boat slightly. This should be taken into account by
RUDDER LIFT modifying the pressure field if it is likely to be a
significant effect compared to the global bending due to
the rig (e.g. on a ketch or a schooner). Much more
Figure 5: A statically determinate set of restraints for a yacht serious for the hull shell itself are the slamming loads
FE model from external wave systems; these need to be analysed
separately and are beyond the scope of this paper.
Once the six degrees of freedom have been restrained,
provided that all the forces on the boat are in equilibrium,
there will be little or no reaction forces at the restraints. 7.6 Limit load cases and restraints
In addition to the equilibrium load case described above,
to check the strength of local parts of the structure it will
usually be necessary to subject the model to a LIMIT
load case. This will almost by definition be a dynamic,
i.e. non-equilibrium situation, so the system of loads and
restraints used for the equilibrium load case will not be
applicable. However, St Venant’s principle[8] states that,
provided the model is restrained sufficiently remotely
from the area of interest, the method of restraint will have
little effect on the local results. Thus it is usually
sufficient simply to “clamp” the model away from the
area of application of load, and ignore the stresses around
the restraints[9].

Figure 6: Strains on a yacht FE model subjected to the 8. PROOF TESTING

W1 equilibrium loadcase, sailing upwind on starboard
tack. The strains on the foredeck are due to the global Demand for ever-higher performance pushes designers
longitudinal bending moment from the rig. The highest towards using smaller safety factors and relying on
strains on the deck are due to transverse compressive testing of the completed structure to check that the
loads from the mast and chainplate bulkheads. The yacht strength is adequate. Such testing allows weaknesses in
is the Baltic 147, designed by Reichel/Pugh and the structure to be detected under controlled conditions,
engineered by SP Technologies. with the minimum risk of consequential damage or injury.
For a series production run of aircraft, it is economically
worthwhile to test a prototype to destruction. For a one-
7.5 Water pressures off yacht, this is generally not the case, so the structure
Restraining the model in this way allows the water forces must be tested to a “proof” load that gives reasonable
to be applied to the model as pressures on the hull confidence in the structure’s ability to support the
surface. With the aid of Computational Fluid Dynamics anticipated loads, without damaging it during testing.
(CFD) analysis, the pressure field can be calculated This raises the question of what load to use for the proof
accurately for a given speed and angle of heel and trim. test. By defining the loads in terms of W1, W2, LIMIT
However, while many yacht development budgets allow and ULTIMATE, the question is more easily answered.
for FE analysis, few can afford CFD. Clearly, to avoid damage to the structure, the proof test
load must certainly be less than the LIMIT load. To
guarantee that the structure can withstand loads in 10. REFERENCES
service, it should be tested to more than the highest load 1 Tye, W. “R.Ae.S Handbook of Aeronautics Structural
it will see, i.e. the W2 load. Given that the factor Principles and Data”, Pitman, 1952.
between these two loads is generally small (of the order
of 1.5) and intended to account for differences between 2 Mosley, W.H, Hulse, R, and Bungey, J.H, “Reinforced
analysis and reality, it is prudent to err towards the lower Concrete Design”, McMillan, 1999.
end of the range and proof test to W2. 3 Barnes, F. “Determination of Design Allowables and
There may be reasons why it is not possible to reach this Validation of Mechanical Properties for Composite
load in a static proof test; for instance there may not be a Materials”, SP Technologies report TR322, June 1989.
suitable way to react the load in a static test. 4 Bruhn, E.F, “Analysis and Design of Flight Vehicle
It is also important to consider the safety implications of Structures” Jacobs (USA), 1973.
a structural failure 5 Gordon, J.E. “Structures”, Penguin, 1978.
during proof testing, particularly where long lengths of 6 Harken Inc. Yacht Fittings Catalogue, 2001.
loaded rope mean that a significant amount of elastic
strain energy is stored in the structure. 7 Sheahan, M “Sailing Rigs and Spars”, JH Haynes,
1990.
8 Timoshenko, S.P. & Goodier, J. N. “Theory of
9. CONCLUSIONS Elasticity”, McGraw Hill, 1970
a) The concentrated loads exerted on yacht structures by 9 Adams, V. and Askenazi, A, “Building Better Products
the rigging can be defined precisely in terms of the with Finite Element Analysis”, OnWord Press, 1999
steady-state component of the load (the W1 value) and
the maximum likely peak value of the load (the W2
value). This avoids the confusion caused by unclear 11. DISCLAIMER
terminology such as “maximum load” or “working load”.
This notice shall be marked on any reproduction of this
b) A system of safety factors can be built into the loads document, in whole or in part. Any reproduction or
used for structural analysis by the Simplified Limit State distribution of this document, in whole or in part, without
Design approach, which is based on the methods used in the prior written consent of SP Technologies is
the civil engineering and aerospace industriel prohibited.
c) The resulting LIMIT load state incorporates the safety This document is intended only as a guide to choosing
factors required to ensure that inaccuracies in the analysis and applying safety factors in the analysis of composite
method, geometric tolerances and other effects will not yacht structures. Following the procedures and using the
cause the structure to be damaged at the highest load it is information contained in this paper does not in any way
likely ever to see in use guarantee that the resulting structure will be fit for
d) The ULTIMATE load state is more severe than the purpose. Furthermore this document provides no
LIMIT state by a factor that should ensure that, even warranty on the suitability or safety of the procedures
considering the statistical spread of load data, the described, nor any guarantee that the loads experienced
likelihood of a catastrophic (i.e. life-threatening) failure by the structure will be less than or equal to the loads
is acceptably low. used for design.
e) In conjunction with the load states defined in this Testing of structures to loads approaching the design
system, the material properties for design must be based failure loads is inherently dangerous and all reasonable
on a statistical analysis of test data to ensure with precautions must be taken to minimise the risk of injury
reasonable confidence that the material in the structure is or secondary damage if the structure or test equipment
at least as strong as assumed for the analysis. This is should fail during testing. This report does not describe
particularly important for composite materials due to the all such suitable safety precautions and it is the
variability inherent in the manufacturing process. responsibility of the person supervising the testing to
ensure that all reasonable precautions have been taken.
f) The Simplified Limit State Design approach allows
combined load cases to be defined for use in Finite As this document cannot be guaranteed to be free from
Element Analysis. errors or omissions, the information contained herein
must be verified independently. SP Technologies
g) Restraining FE models of yachts without adding assumes no liability or responsibility to any person or
stiffness or causing spurious stresses is not organisation for direct or indirect damages resulting from
straightforward. A system of statically determinate the use of any information contained in this paper.
restraints at the rigging attachments has been proposed to
avoid these pitfalls.
h) The completed structure should, wherever possible, be
“proof tested” to a suitable load. Usually this is the W2
load.