Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/353594772

An engineering method to design daggerboards and rudders for small racing


yachts

Article  in  Ocean Engineering · July 2021


DOI: 10.1016/j.oceaneng.2021.108746

CITATIONS READS
0 380

4 authors, including:

Stefano Castegnaro
Area Impianti Spa
22 PUBLICATIONS   84 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

IPANEMA: Inlet PArticle Separator Numerical ExperiMental Assessment View project

Fan design View project

All content following this page was uploaded by Stefano Castegnaro on 16 August 2021.

The user has requested enhancement of the downloaded file.


Ocean Engineering
An International Journal of Research and Development ISSN: 0029-8018

An engineering method to design daggerboards and


rudders for small racing yachts
S. Castegnaro, C. Battisti, M. Poli, A.Lazzaretto
https://doi.org/10.1016/j.oceaneng.2021.108746

free access to the full article, until September 18th, 2021


https://authors.elsevier.com/a/1dUvw6nh6z5XN

NOTICE: this is the author’s version of the work that was accepted for publication. Changes resulting
from the publishing process, such as peer review, editing, corrections, structural formatting, and other
quality control mechanisms may not be reflected in this document.
An engineering method to design daggerboards and
rudders for small racing yachts
S. Castegnaro, C. Battisti, M. Poli, A.Lazzaretto
Highlights
An Engineering Method to Design Daggerboards and Rudders for Small Racing Yachts
S. Castegnaro,C. Battisti,M. Poli,A. Lazzaretto

◦ The different appendages of dinghies indicate the lack of a comparative design method
◦ A VPP compares the performance of different appendages, providing design indications

◦ The optimum design is identified as best trade-off of upwind and leeward conditions
◦ Elliptic and Rectangular foils with same draft but different AR perform similarly
◦ The design of the rudder requires specific insights due to its complex fluid-dynamics
An Engineering Method to Design Daggerboards and Rudders for
Small Racing Yachts
S. Castegnaroa,∗ , C. Battistib , M. Polic and A. Lazzarettod
a von Karman Institute for Fluid Dynamics, Waterloosesteenweg 72, B-1640 Sint-Genesius-Rode, Belgium
b Fincantieri,
Passeggio Sant’ Andrea 6/a, Trieste 34123, Italy
c Free Consultant, Munich, Germany
d Department of Industrial Engineering, University of Padova, via Venezia 1, Padova 35131, Italy

ARTICLE INFO ABSTRACT


Keywords: The variety of shapes and sizes among daggerboards and rudders of small racing yachts indicates the
Daggerboard Design lack of a comparative design method and that some aspects may be neglected in the designs currently
Centreboard Design proposed. In particular, a literature survey showed that: i) the conventional design approaches are
Rudder Design exclusively based on upwind conditions, disregarding any downwind contribution to the race, ii) the
Appendage Design detrimental effects related to the typically low Reynolds numbers involved are marginally considered,
Small Crafts and iii) there are ambiguous indications about sweep angle effects on foil performance. Based on
Small Yachts a Velocity Prediction Program, a design approach compares the different appendage planforms of
Sweep Angle small racing yachts in terms of size and shape, identifying the ones minimizing the overall racing
time, sum of upwind and downwind contributions. The method is applied to a 4.6 m skiff and the
performance of the daggerboards analysed are compared with xflr5 simulations, validated against
experimental tests. The results quantify the time gains of longer foils and the similar performance
of elliptic and rectangular geometries with equal drafts but different aspect ratios. Swept foils are
investigated as well, showing lower lift curve slopes compared to straight geometries for the high
aspect ratios typically used on small yachts.

Nomenclature 𝐶𝐿 : Lift coefficient of the entire foil (or of the sail) [-]

AWA: Apparent wind angle (with respect to the boat motion) [°] 𝐶𝐷 : Drag coefficient of the entire foil (or of the sail)

AWS: Apparent Wind Speed [m/s] 𝐶𝐷𝑖 : Drag coefficient of the entire foil induced by the lift

awa’: Apparent wind angle (with respect to the boat longitudinal 𝑑: Appendage draft (or semi-span of the symmetric wing made
centreline) [°] by the daggerboard with its mirrored image)

𝐴𝑅: Aspect ratio, 𝑑 2 ∕𝑆 [-] D: Drag [N]

𝐴1𝑟𝑒 : First coefficient for a rectangular foil (see [1]) 𝐹 𝑟: Froude number = √
𝐵𝑆
[-]
𝑔⋅𝐿𝑊 𝐿

𝑎: Lift curve slope [1/rad] 𝐹𝑡′ , 𝐹ℎ′ Driving and heeling forces along the boat centreline and
BS: Boat speed [m/s] transversal direction [N]

𝑐𝑡 : Tip chord length [m] 𝐹𝑃 : Driving (propulsive) force along the boat motion [N]

𝑐𝑟 : Root chord length [m] 𝐹𝑆 : Side force orthogonal to the boat motion [N]

𝑐𝑚 : Mean chord length, 𝑆∕𝑑 [m] 𝑓 : Parameter for computation of the lift curve slope of tapered
foils (see [2])
𝐶𝑡 : Propulsive coefficient of sails [-]
𝑓1 , 𝑓2 , 𝑓3 , 𝑓4 : functions
𝐶ℎ : Heeling (side) force coefficient of sails [-]
g: Acceleration of gravity, 9.806 [𝑚∕𝑠2 ]
𝐶𝑙 : Lift coefficient of the foil section (infinite AR) [-]
𝐻𝑀: Heeling Moment [Nm]
𝐶𝑙𝑚𝑎𝑥 : Maximum lift coefficient of the foil section (infinite AR)
[-] 𝐻1∕3 : Significant wave height [m]

𝐶𝑑 : Drag coefficient of the foil section (infinite AR) [-] I: Moment of inertia [𝑚𝑚4 ]

𝐶𝑑𝑚𝑖𝑛 : Minimum drag coefficient (at zero lift) of the foil section 𝑙𝑅𝑒𝑔 : Distance between the upwind-downwind buoys during a
(infinite AR) [-] regatta

∗ Corresponding author L: Lift [N]


stef.castegnaro@gmail.com (S. Castegnaro)
ORCID (s): 0000-0001-5142-8742 (S. Castegnaro) 𝐿𝑂𝐴 Length overall [m]

Castegnaro et al.: Preprint submitted to Elsevier Page 1 of 26


An Engingeering Method to Design Daggerboards and Rudders

𝐿𝑊 𝐿 : Length waterline [m] Subscripts


i: Generic iteration
𝑀𝑎: Mach number [-]
BH: Bare hull
𝑁𝑐𝑟𝑖𝑡 : xfoil/xflr5 parameter [-]

𝑅𝑒: Reynolds number


𝜌𝑤 ⋅𝐵𝑆⋅𝑐𝑚 DB: Daggerboard
𝜇

𝑅𝑀: Righting Moment [Nm] R: Rudder

S: Appendage surface [𝑚2 ] 𝜈: Viscous

TWA: True wind angle (with respect to the boat motion) [°] opt: Optimal

TWS: True Wind Speed [m/s] 2D: Referred to the foil 2D section (i.e., infinite AR)
twa’: True wind angle (with respect to the boat longitudinal cen- TOT: Total
treline) [°]
S: Referring to the sails
𝑇1 : Wave period [s]
UW: Upwind
𝑦𝑚𝑎𝑥 : Distance between the daggerboard axis and the point of
maximum thickness on the root cross section [mm] DW: Downwind
𝑉 𝑀𝐺: Velocity Made Good (i.e., projection of 𝐵𝑆 on the 𝑇 𝑊 𝑆 el: Elliptic
direction) [m/s]
re: Rectangular
𝑉 𝑃 𝑃 : Velocity Prediction Program
𝑘𝑦𝑦 : Longitudinal radius of gyration (≈ 0.25 ⋅ 𝐿𝑂𝐴 - see [3]) [m] tap: Tapered

▽𝑐 : Volume displacement of the hull [𝑚3 ] 50: Mid-chord length

a, b: Coefficients for the computation of the added resistance


√ in

waves, tabulated (see [3, p.95]) as function of 𝑇1 = 𝑇1 𝑔∕𝐿𝑊 𝐿 1. Introduction
and of 𝐹 𝑟 for the different courses.
This paper presents an engineering method to address the
W: Crew weight [kg] design of the appendages (daggerboard and rudder) of small
racing yachts, providing indications to identify: i) the foil
k: Ratio between root chord length and daggerboard draft = 𝑐𝑟 ∕𝑑
surfaces (in 𝑚2 ) and ii) the planform shapes and drafts which
[-]
achieve high performance during a regatta with upwind and
𝑘𝑣 : Coefficient in Eq. 16 downwind legs.
𝑧: Vertical coordinate [m] In spite of the current interest for flying sailboats (i.e.,
boats with underwater lifting appendages that sail with the
Greek Symbols hull above the water), it seems reasonable that the vast ma-
𝛼: Angle of Attack [°] jority of future recreational racing yachts will remain of the
Δ: Difference conventional type. With this regard, small racing sailboats
as dinghies and skiffs (i.e., recreational crafts whose length
𝛾: Leeway angle [°] overall is typically around 4-6 m and with a crew of two-
𝜌𝑎𝑖𝑟 : Air mass density, 1.2 [𝑘𝑔∕𝑚3 ] three persons) are addressed in this paper. Historically, these
boats sail and race since several decades (see e.g., [4]); how-
𝜌𝑤 : Water mass density, 1020.0 [𝑘𝑔∕𝑚3 ] ever, the design of their appendages has always been a rather
𝜇: Water dynamic viscosity, 8.90 ⋅ 10−4 [𝑃 𝑎 ⋅ 𝑠] vexed topic and it is a typical subject of discussion among
yacht designers or in the nautical clubs. As a matter of fact, it
Φ: Downwash angle [rad]
is common finding dinghies with similar dimensions and sail
𝜆: Tip to chord ratio 𝑐𝑡 ∕𝑐𝑟 [-] surfaces, but with different daggerboards and rudders. The
natural question arises: which foil shape will be the fastest
Λ: Sweep angle (referred to the 0.25 of the chord length posi-
during a race? The technical literature on nautical design
tion) [°]
does not provide a clear reply nor all the indications, as well
𝜏: Rectangular foil coefficient, see [1] summarized by Trower [5, p.105]: "The wide range of plan-
𝜎𝑎𝑑𝑚 : Admissible stress [MPa] forms used for keels also supports the argument that no one
has all the answers..".
𝛿, 𝜙: Angles [°] The authors of this paper experienced the lack of a unified
𝛿𝐵𝐿 : Boundary layer thickness [m] comparative design method for foils in several ways, dur-
ing more than 10 years of participation to the 1001Vela Cup
𝜇𝑟𝑒 : Geometric coefficient for the rectangular foil (= 2𝜋 ⋅ 𝑐𝑟 ∕8𝑑) competition (http://www.1001velacup.eu/ - a race between
[-]

Castegnaro et al.: Preprint submitted to Elsevier Page 2 of 26


An Engingeering Method to Design Daggerboards and Rudders

Figure 1: Daggerboard/Centreboard/Keel and Rudder surfaces against the Sail plane area
for different small sailboats from a benchmarking analysis (see Appendix A). Total Area
represents the corresponding sum for the two appendages of each boat.

skiff prototypes designed, manufactured and sailed by uni- design solely on the upwind conditions (where foils act sim-
versity students). The first macro-indication came from ob- ilarly to the wings of an airplane to compensate the side force
serving the different competitors: although constrained by on the sails), neglecting or giving little consideration to the
the same design rules, the boats are generally presenting dif- downwind navigation. However, the downwind leg still rep-
ferent appendages. A further confirmation came from a bench- resents around half of the race, during which the side force
marking (reported in Appendix A) performed on 20 com- compensation is not needed anymore and the daggerboard
mercial dinghies and skiffs with similar dimensions to the shall produce as less drag as possible to avoid slowing down
ones of the 1001Vela. Some data from this analysis are shown the craft. Therefore, in principle, it seems reasonable that
in Fig. 1, which presents the daggerboard (or centreboard/keel, the optimum daggerboard design shall represent a trade-off
according to the case) and rudder surfaces of each boat against between both the upwind and downwind conditions of the
the corresponding sail plane area. A first look at the graph race.
of Fig. 1 indicates an increasing trend of the foil surfaces In second instance, in spite of the rather unanimous aware-
against the sail areas, but with a significant scatter among ness in the literature (e.g., [5, 6, 7, 8, 11]) about the detri-
the different data. In spite of the crudeness of this survey mental effects related to typically low-to-moderate Reynolds
(further analyses should evaluate whether such differences numbers (i.e., 𝑅𝑒 < 3 ⋅ 106 ) on foil performance, no indi-
can be explained in terms of different design choices), at the cations are provided on how systematically accounting for
same time several planform shapes can be observed among these detriments directly in the preliminary sizing of the foils.
the foils of these boats (Fig. 15 in Appendix A): elliptic, This latter aspect is especially relevant for skiffs and dinghies,
tapered, rectangular geometries, as well as different sweep because of the relatively small chord lengths of their ap-
angles, have been adopted during the years by different de- pendages. At last, some disagreement was found in the lit-
signers. erature about the performance gain/loss of swept foils with
Seeking for a unified design methods for the appendages, respect to straight ones.
a research was performed in the technical literature on yacht From a broader point of view, in spite of the different
design [3, 4, 5, 6, 7, 8, 9, 10, 11]; in first instance, it was successful foil designs proposed in the technical literature, a
concluded that most of these methods base the appendage unified approach which compares the basic design choices in

Castegnaro et al.: Preprint submitted to Elsevier Page 3 of 26


An Engingeering Method to Design Daggerboards and Rudders

terms of the fundamental principles of the fluid-dynamic of shall be remarked that several successful and refined foil de-
sailing to minimize the racing time is not available. Indeed, signs exist nowadays and the approach presented here does
it is rather common performing a preliminary design (i.e., not have the ambition of providing optimized geometries.
deciding the foil surface or draft) from benchmarkings on The objective is rather providing a solid procedure to com-
similar crafts (see the successful design [11], for instance). pare the basic design parameters which define the appendage
On the other hand, refined numerical tools exist in support of planforms of small racing yachts, accounting for aspects gen-
the foil design exercise (e.g., [12, 13]), but their use is gener- erally neglected as the contribution of the downwind part of
ally limited to the highest levels of sail competition, like the the race and the effects related to the Reynolds number.
America’s Cup regatta. According to the authors’ experi-
ence in fact, the small teams normally involved in designing 2. Method
a new yacht (a commercial or a prototype one) and of her ap-
pendages seldom have the resources for including numerical The method used to identify the optimum planforms and
optimization campaigns during the design phase (especially sizes of the daggerboard-rudder combination is described
in case of small sailboats, being the fees proportioned to the here in a step-by-step procedure. The method is a VPP for
boat dimensions). the boat with the different appendage combinations, that is
It thus follows the need for a quick, simple, but still ef- based on computing the balance between forces (the side
fective design method, to identify the basic shape and size of force on the sails against the lift of the daggerboard, and the
the daggerboards and rudders of small racing yachts. With propulsive force against the drag of the system). Some pre-
this regard, the method developed by the authors is presented liminary tasks, requiring familiarity with the fields of sailing
in this paper: for a given craft, it is based on the performance fluid-dynamics and nautical design, are needed to correctly
evaluation of different daggerboard-rudder combinations by implement the VPP algorithm: in particular, the availability
a Velocity Prediction Program [7, 14], during the upwind of the bare hull characteristics and of the sail polars, as well
and the downwind legs of the regatta (assuming a windward- as the awareness of the expected wind and sea conditions at
leeward race). In the VPP a method developed by Jacobs and the race field, are necessary prerequisites. The method ulti-
Sherman [15] is integrated to estimate the polars of the foil mately provides an estimation of the overall time 𝑡𝑇 𝑂𝑇 (sum
sections at different Re, thus systematically accounting for of the times during the upwind and downwind legs) and this
the detriments on the foil performance. parameter is used to rank the different appendage combina-
The results of the method are summarized into plots which tions: the quickest is the best. Berman [17] likely makes use
identify the daggerboard and rudder geometries providing of a similar approach to compare the performance of keels
the best upwind-downwind trade-off during the regatta (i.e., against daggerboards for a sailing catamaran.
the minimum overall racing time) for the expected wind and The hypotheses on which the method is based are those
sea conditions. Since the results are based on the estima- typical of appendage design for small racing boats (see for
tion of the foil performances, the prediction capability of the instance [5, 6, 11]) and are here summarized:
computations are compared with simulations performed on • the boat is assumed to be in a perfectly flat position
the same geometries with the xflr5 code [16]. In turn, the (i.e. no heeling);
xflr5 analyses have been compared with experimental data
• the daggerboard lift counteracts entirely the side force
on two wings at 𝑅𝑒 = 2 ⋅ 106 , to assess the accuracy of the
on the sails (i.e., there is no contribution from the rud-
code at low-to-moderate Reynolds. This numerical tool is
der, which is instead designed primarily for steering
used here in order to suggest a relatively easy and accessible
the boat);
verification for anyone who is facing the task of appendage
design for skiffs and dinghies. • the hull represents a symmetry plane for the dagger-
The article is organized as follows: in section 2 the entire board (the mirroring image concept, see e.g., [3, 6,
architecture of the method is reported in a descriptive way 10]), whose equivalent aspect ratio (𝐴𝑅𝐷𝐵 ) is thus
(leaving its mathematical development, the equations, and a 2
twice the geometric one (= 𝑆𝑑 , with 𝑑 and 𝑆𝐷𝐵 being
numerical example in Appendix B for sake of readability); in 𝐷𝐵
respectively the draft and surface of the daggerboard);
section 3 the procedure is applied to the case-study of a 4.6
m skiff and the results in terms of daggerboard performance • the rudder is placed on the boat stern (transom hung),
are compared with the analogous parameters obtained from right over the length overall of the craft;
xflr5 analyses. In section 4 indications on peculiar aspects
of rudder design, as well as for structural and stability veri- • the wave drag or residual drag of the appendages are
fications, are provided. Finally, in section 5 further consid- neglected;
erations concerning the selection of the foil section and the • the effect of the leeway angle (𝛾) on the resistance of
sweep angle are reported together with a discussion about the bare hull (𝐷𝐵𝐻 ) is assumed to be negligible, thus
the limitation of the work. 𝐷𝐵𝐻 is only function of the boat speed (𝐵𝑆);
The paper is primarily intended for designers of small yachts:
the knowledge of the fluid-dynamics of sailing and of the re- • when the rudder is not steering the boat, it is posi-
lated design/analysis tools are considered prerequisites. It tioned at an angle of attack around the zero lift posi-
tion (thus, its drag is exclusively the viscous one).

Castegnaro et al.: Preprint submitted to Elsevier Page 4 of 26


An Engingeering Method to Design Daggerboards and Rudders

The approaches are thus different for the daggerboard and cr


the rudder:
Daggerboard - It is designed to compensate the entire
side force from the sails and its performance are evalu-
ated according to the conventional fluid-dynamics equa-
tions of wings. Geometries corresponding to differ-
ent: i) drafts 𝑑 (i.e., wing semispans), ii) Aspect Ratio ct =crλ
𝐴𝑅𝐷𝐵 , and iii) planform shapes (e.g. elliptic against
rectangular) are compared. For a given draft and plan-
0.1 0.5 1.0
form shape, the aspect ratio is systematically varied k = cr /d
modifying the ratio 𝑘 = 𝑐𝑟 ∕𝑑 between the root chord
length and the daggerboard draft, as shown in Fig. 2. Figure 2: The 𝑘 ratio relates the root chord length 𝑐𝑟 with the
For a given planform shape, this allows to express the draft 𝑑, thus identifying the entire geometry of the dagger-
main foil geometrical features (surface size, aspect ra- board for conventional straight elliptic, 1/4 elliptic, rectangu-
lar, and tapered shapes. For the tapered one, the tip-to-chord
tio) as function of the draft 𝑑.
ratio 𝜆 is required as well.
Rudder - It needs to respond to two conflicting require-
ments: low drag during navigation on one hand, and
the capability of delaying as much as possible the stall, 1. Several inputs are needed to start:
on the other (with this latter being somehow the pre-
• The race environmental conditions, in terms of:
eminent one [5, 7, 8]).
expected True Wind Speed (TWS), leg length (𝑙𝑅𝑒𝑔
The rudder is sized according to an empirical rule, as - i.e., the distance between the upwind and down-
a proportion of the lateral underwater plane (see [5]), wind buoys), expected wave period (𝑇1 ), and sig-
targeting primarily the capability of steering the boat nificant wave height (𝐻1∕3 );
(this approach is described in detail in Sec. 4.1). • The hull characteristics: the length overall and
Apart from the diverse operations, the different approaches length at waterline (𝐿𝑂𝐴 and 𝐿𝑊 𝐿 ), the Bare
are needed because on dinghies the rudder is placed in line Hull resistance curve 𝐷𝐵𝐻 = 𝑓1 (𝐵𝑆) as ob-
behind the daggerboard, thus facing the complex flow char- tained by software simulation (e.g., MaxSurf) or
acterized by the wake and the downwash of the first foil. by towing tank tests, the hydrostatic wet surface
Similar distinctions are adopted by other authors as well (see (𝑆𝐵𝐻 ), and the volume displacement of the hull
for instance [5, 11]). (▽𝑐 );
• the upwind and downwind sail dimensions (𝑆𝑆𝑈 𝑊
2.1. VPP Architecture and 𝑆𝑆𝐷𝑊 , respectively), which are needed to
The architecture of the VPP is presented here in a se- compute the propulsive and side forces;
quential form and with the support of the flow chart in Fig. • the polar characteristics of the sails (both upwind
3; for sake of readability, the structure is presented (as much and downwind), to provide the force coefficients
as possible) in a descriptive way, reporting the entire mathe- which will be used to compute the propulsive
matical development, the equations and a numerical example and side forces (𝐹𝑝 and 𝐹𝑠 , respectively) accord-
in Appendix B. ing to the boat heading;
The performance estimation is based on an iterative compu-
tation which assumes a boat speed (𝐵𝑆) of first guess for the • the polar characteristics of the 2D foil section
combination of appendages under analysis: the unbalance at several Reynolds numbers, to account for the
between the propulsive force (𝐹𝑝 ) and the total drag (𝐷𝑇 𝑂𝑇 performance variations with Re. With this re-
- including hull, daggerboard and rudder) is then used to cor- gard, experimental databases for symmetric air-
rect 𝐵𝑆. When the computation converges, it is possible to foils can be obtained from the reports [15, 18,
compute the projection of the boat speed on the wind di- 19] or could be built with the open-source XFOIL
rection (i.e., the Velocity Made Good VMG) and thus, the software [20].
resulting time (𝑡𝑈 𝑊 or 𝑡𝐷𝑊 for the upwind or downwind This input requires the selection of specific foil
cases, respectively) to cover the distance between the two sections for the daggerboard and the rudder (in-
buoys. When the computation has been applied to both the dications in this respect are specifically addressed
upwind and downwind legs, it is finally possible to compute in Sec. 5.1).
the overall time 𝑡𝑇 𝑂𝑇 = 𝑡𝑈 𝑊 + 𝑡𝐷𝑊 . 2. In step 2 the design Parameters for the daggerboard
Referring to Fig. 3, each block of the computation is pre- shall be given as input as well, and then modified in
sented in the following list (see the corresponding numbers). successive computations to assess different geometries.
Taking familiarity with the flow-chart of Fig. 3 and with the These parameters are: the daggerboard draft (or semi-
Nomenclature is recommended before proceeding further. span, in the wing-daggerboard analogy) 𝑑 selected,

Castegnaro et al.: Preprint submitted to Elsevier Page 5 of 26


An Engingeering Method to Design Daggerboards and Rudders

1. INPUTS:
10. aDB=f(planform; 16.
o TWS [m/s]; lReg [m];
o DBH resistance curve [N]; SBH [m2] AR)
DR @ 0° [N]
o LOA, LWL [m], Vc [m3]
o SS_UW; SS_DW [m2]
o Sail coefficients = f(awa’) 11. CL=f(aDB;Fs)
o awa’ [°]
o Hydrofoil section polars @ several Re
o Wave period T1 and sign. heigth H1/3
12. CDi=f(AR; CL); 0;
[°]

2. DAGGERBOARD
BS INPUT DESIGN
PARAMETERS: 13. Cd=f(CL; Re)

 o draft d [m]

o Range of k=cr/d 14. CD = CDi + Cd ; DDB


(e.g., 0.1<k<0.7)

o Planform shape
17. DAW = f(a,b,Fr,T1,H1/3)

3. 5. VELOCITY 18. DTOT= DBH +DDB +DR (+DAW)


TRIANGLES: AWS
DBH=f(BS)
[m/s]; TWA [°] ki
[N] SBH
6. PROPULSIVE &
NO 19.
SIDE FORCES: Fp,
Fs [N] DTOT = Fp ?

7. DAGGERBOARD
8. YES
ReDB=f(BS,
PARAMETERS:
BS 
d, ki) 
ARDB; SDB [m2]; cm [m]
[°]

15. RUDDER 21. VMG=f(BS,


9. SURFACE: TWA)
a2D=f(section,
2
ReDB) SR [m ];
t

22.
k

t=f(d,k,planform) [s]

Figure 3: The VPP algorithm architecture for the performance estimation of different
appendage combinations: for a given draft and planform shape, it is applied to different
𝑐𝑟 ∕𝑑 ratios to account for different 𝐴𝑅𝐷𝐵 , with the convergence criterion on the balance
between propulsive force 𝐹𝑡 and resistance 𝐷𝑇 𝑂𝑇 . The procedure is applied on the upwind
and downwind conditions to compute the overall time: 𝑡𝑇 𝑂𝑇 = 𝑡𝑈 𝑊 + 𝑡𝐷𝑊 .

the range of 𝑘 ratios to assess during the computation, Each computation is performed on a single value of 𝑘𝑖
and the generic planform shape (e.g., 1/4 of ellipse or and, once converged, it is then repeated with the 𝑘𝑖+1
rectangular). All the geometrical parameters of the one. When the computations on the entire 𝑘 range are
daggerboard (surface 𝑆𝐷𝐵 , aspect ratio 𝐴𝑅𝐷𝐵 , and completed, the results are recorded and the procedure
chord length distribution) are so defined exclusively applied again for different daggerboard drafts 𝑑 and
in terms of 𝑘 and 𝑑 for a given planform (see table 7 planforms for comparison.
in Appendix B for the mathematical formulas).
3. Once the boat speed 𝐵𝑆 is imposed at the beginning

Castegnaro et al.: Preprint submitted to Elsevier Page 6 of 26


An Engingeering Method to Design Daggerboards and Rudders
aw
a'

γ AWS 1.05
TWS δ
BS 1
VMG
AW

a2D/2π
A 0.95

AWA
BS
φ

h h 0.9

awa'
VMG

γ
φ 0.85 0012
TW
A
64012
0.8
δ 0 2 4 6 8 10
Re x 10
6

A
TWS TW
AWS (a) The variation of the lift curve slope (with respect to the theoret-
ical flat plate value 2𝜋) with the Reynolds number.
Figure 4: The Velocity Triangles to compute the forces
on the sails.
−3
x 10
7

of the iteration, 𝐷𝐵𝐻 is computed from the resistance


curve of the bare hull (which can be approximated by 6

min
an interpolating polynomial, for instance);
Cd 5
4. A leeway angle 𝛾 of first guess is assumed (typically,
𝛾1 = 0°); 0012
64012
4
5. The velocity triangles on the sails (Fig. 4) are re- 0 2 4 6 8 10
Re 6
solved with the 𝐵𝑆 and 𝛾 resulting from the previous x 10
iteration, finally providing the Apparent Wind Speed
(𝐴𝑊 𝑆). (b) The decreasing trend with 𝑅𝑒 of the drag coefficient at zero lift.

6. The propulsive and side forces 𝐹𝑝 and 𝐹𝑠 are com-


puted projecting the actions on the sails on the direc- 1.6
tion of the motion (𝐵𝑆) and its perpendicular. Ac-
cording to the heading under analysis, upwind or down- 1.4
wind sail dimensions and related force coefficients are
max

used. 1.2
Cl

7. The Daggerboard geometrical parameters are computed 1


(𝑆𝐷𝐵 , 𝐴𝑅𝐷𝐵 , and the mean chord 𝑐𝑚 ) from the inputs 0012
64012
provided at step nr. 2. As previously reported, the ef- 0.8
0 2 4 6 8 10
fective aspect ratio for the performance computation Re x 10
6

of the daggerboard is doubled, according to the mir-


roring plane concept induced by the hull [3, 6, 10].
(c) The trend of the the maximum lift coefficient with 𝑅𝑒.
8. The daggerboard Reynolds number is computed refer-
ring to the mean chord length 𝑐𝑚 . Since 𝑐𝑚 ≈ 𝑐𝑟 = Figure 5: Variation of the foil section performances with
𝑘 ⋅ 𝑑, for a given 𝑑 the Reynolds number 𝑅𝑒𝐷𝐵 is ap- the Reynolds number for two NACA symmetric profiles
proximately proportional to the 𝑘𝑖 of that particular [18].
iteration (thus, 𝑅𝑒𝐷𝐵 increases linearly with 𝑘 - see
Fig. 2). Conversely, at equal 𝑘, 𝑅𝑒 increases with 𝑑.
thickening of the boundary layer on the suction side
9. For the daggerboard, the lift curve slope pertaining to of the foil section with decreasing 𝑅𝑒 (a well known
the selected foil section is identified according to the phenomenon in several technical disciplines, as turbo-
Reynolds number: 𝑎2𝐷 = 𝑓2 (𝑅𝑒𝐷𝐵 ), with 𝑓2 being machinery design - see e.g., [21, p.71]).
an interpolation function fitting the data. As shown
in Fig. 5a for the 0012 and 64012 NACA sections 10. The lift curve slope of the entire three-dimensional
[18], the 𝑎2𝐷 parameter features important variations daggerboard (𝑎𝐷𝐵 ) is computed according to the well-
for 𝑅𝑒 < 3⋅106 , which is the range of interest for most known equations coming from the aerodynamics of
of the appendages [5, 8]. This behaviour is due to the wings. These equations (cf. [2, 1, 22, 23]) allow to

Castegnaro et al.: Preprint submitted to Elsevier Page 7 of 26


An Engingeering Method to Design Daggerboards and Rudders

account for the different planforms (elliptic, tapered, analytic approach as the one used for the daggerboard
rectangular), as well as for the sweep angle Λ. also for the rudder. With this respect, Reynolds Av-
eraged Navier Stokes (RANS) simulations performed
11. The daggerboard lift coefficient (𝐶𝐿𝐷𝐵 ) required to bal- in [26] on six daggerboard-rudder combinations for a
ance the side force 𝐹𝑠 on the sails is computed as: skiff sailing at a given leeway angle showed a decrease
𝐹𝑠 of lift for the rudder around Δ𝐿𝑅 = 31 − 34% with re-
𝐶𝐿𝐷𝐵 = (1) spect to the same geometry in isolated conditions, and
0.5𝜌𝑤 ⋅ 𝐵𝑆 2 ⋅ 𝑆𝐷𝐵 a lower decrease of drag (Δ𝐷𝑅 = 3 − 11%). This
where 𝜌𝑤 is the mass density of water. By knowing aspect is recalled in Sec. 4.1.
𝐶𝐿𝐷𝐵 and the lift curve slope 𝑎𝐷𝐵 just computed, the The positioning of the daggerboard with respect to the
daggerboard angle of attack 𝛼 (which will ultimately rudder and the centre of force on the sails [5, 7]) does
be equal to the leeway angle) can be calculated as well. not affect the performance estimation. Nonetheless,
the boat centering has to be verified (see Sec. 4) to
12. With the lift coefficient obtained at the previous step, achieve the sailing balance before finalizing the foil
the unviscid drag coefficient 𝐶𝐷𝑖 (related to the in- design.
duced velocity, and thus to the lift) is computed.
16. Since the rudder is sized for maneuvering and not to
13. Unless the foil section polars at different 𝑅𝑒 are di- contribute in resisting leeway, at this step the sailing
rectly integrated within the computation (as for in- balance (i.e., the equilibrium between turning moments
stance in the Xfoil-based method presented in [24] for along the yaw axis of the boat) is assumed to be achieved.
fan blade design), the viscous drag coefficient 𝐶𝑑 can This involves that the drag of the rudder during navi-
be computed at the corresponding incidence angle ac- gation (unless if used to correct the trajectory) is only
cording to the generalized section polar proposed by the viscous one around the zero-lift operation:
Jacobs and Sherman [15], which allows estimating the
polar of the generic 2D section at different Reynolds 1
𝐷𝑅 = 𝜌 𝐶 ⋅ 𝐵𝑆 2 ⋅ 𝑆𝑅 (5)
numbers. This approach (described in Appendix B) 2 𝑤 𝑑𝑚𝑖𝑛
requires the trends with Reynolds of: i) the minimum
where 𝐶𝑑𝑚𝑖𝑛 is function of the Reynolds number and
drag coefficient 𝐶𝑑𝑚𝑖𝑛 , and ii) the maximum lift coef-
of the foil section selected for the rudder. According
ficient 𝐶𝑙𝑚𝑎𝑥 of the selected foil section (see Fig.s 5b
to the data reported in [26], the decrease of rudder
and 5c for the NACA 0012 and 64012).
drag caused by the daggerboard wake and downwash
14. The overall daggerboard drag is computed as sum of is Δ𝐷𝑅 < 11%; this makes the computation of 𝐷𝑅
the viscous and non-viscous contributions: with Eq. 5 an acceptable approximation even for rud-
ders positioned in line behind the daggerboard.
𝐶𝐷𝐷𝐵 = 𝐶𝑑 + 𝐶𝐷𝑖 (2)
17. Only during the upwind navigation, an additional drag
1 is caused by the boat interaction with the waves nat-
𝐷𝐷𝐵 = 𝜌𝑤 𝐶𝐷𝐷𝐵 ⋅ 𝐵𝑆 2 ⋅ 𝑆𝐷𝐵 (3)
2 urally existing on the race field. This is called added
resistance in waves 𝐷𝐴𝑊 and it is modelled in terms
15. As anticipated, the rudder is sized according to a rule of the boat characteristics and of the waving system
of thumb suggested by Trower for dinghies [5, p.113]: (see [3, 27]).
its surface 𝑆𝑅 is assumed to be equal to the 15% of the
underwater lateral plane (i.e., the lateral projection of 18. At the given 𝐵𝑆, the entire drag of the hull with the
the underwater surface 𝑆𝐵𝐻𝑙𝑎𝑡 + 𝑆𝐷𝐵 + 𝑆𝑅 ), giving: appendages (𝐷𝑇 𝑂𝑇 = 𝑓 (𝐵𝑆)) is then computed as

𝑆𝑅 ≈ 0.176 ⋅ (𝑆𝐵𝐻𝑙𝑎𝑡 + 𝑆𝐷𝐵 ) (4) 𝐷𝑇 𝑂𝑇 = 𝐷𝐵𝐻 + 𝐷𝐷𝐵 + 𝐷𝑅 (+𝐷𝐴𝑊 ) (6)

The reasons for using the empirical relation 4 are mainly where 𝐷𝐴𝑊 is included only for the upwind leg.
two. First, we experienced that the rudder surfaces ob-
19. Finally, the initially guessed boat speed value 𝐵𝑆 is
tained from Eq. 4 allow an acceptable compromise
verified comparing the propulsive force on the sails
between the navigation conditions and the need for
𝐹𝑝 with the total drag 𝐷𝑇 𝑂𝑇 . In particular, 𝐵𝑆 is in-
effective maneuvering capabilities at low boat speeds
creased if
(e.g., approaching an harbour or at the start-line of the
regatta). On the contrary, more complex approaches 𝐷𝑇 𝑂𝑇 < 𝐹𝑝
resulted into undersized rudders, with poor effective-
ness at low 𝐵𝑆. In second instance, the fact that dur- or decreased if
ing navigation the rudder is immersed into the wake
and downwash of the daggerboard (see [25] for con- 𝐷𝑇 𝑂𝑇 > 𝐹𝑝
ventional yachts) calls into question the accuracy of an

Castegnaro et al.: Preprint submitted to Elsevier Page 8 of 26


An Engingeering Method to Design Daggerboards and Rudders

until converging to a value such that 𝐷𝑇 𝑂𝑇 = 𝐹𝑝 . At Table 1


each new iteration the leeway angle previously com- Data and polynomial functions used in the computations. The
puted returns to step 4, updating so the computation upwind sail force coefficients have been taken from the numeri-
of the velocity triangles. cal aerodynamic analyses of the skiff reported in [28] , whereas
the downwind ones from [3]. 𝑓4 pertains to the method by Ja-
20. When the iterations are converged, the daggerboard cobs and Sherman [15], reported in the Appendix B.
angle of attack 𝛼 actually is the leeway angle of the
boat 𝛾. TWS 5.0 m/s 𝑙𝑅𝑒𝑔 2000 m
𝜌𝑎 1.2 𝑘𝑔∕𝑚3 𝜌𝑤 1020.0 𝑘𝑔∕𝑚3
21. The Velocity Made Good VMG is computed as: 𝑆𝑆𝑈 𝑊 16.34 𝑚2 𝑆𝑆𝐷𝑊 28.81 𝑚2
𝑎𝑤𝑎′𝑈 𝑊 25° 𝑎𝑤𝑎′𝐷𝑊 100°
𝑉 𝑀𝐺 = 𝐵𝑆 ⋅ |𝑐𝑜𝑠(𝑇 𝑊 𝐴)| (7) 𝐶𝑡𝑈 𝑊 0.38 𝐶𝑡𝐷𝑊 1.06
𝐶ℎ𝑈 𝑊 1.15 𝐶ℎ𝐷𝑊 1.12
22. The times 𝑡𝑈 𝑊 and 𝑡𝐷𝑊 are finally calculated for the 𝐿𝑊 𝐿 4.6 m ▽𝑐 0.242 𝑚3

𝐻1∕3 0.3 m 𝑇1 2.5 s
upwind and downwind legs respectively, dividing the
distance between bouys 𝑙𝑅𝑒𝑔 for the VMG:
𝑓1 = 0.5554 ⋅ 𝐵𝑆 6 − 6.3058 ⋅ 𝐵𝑆 5 + 22.5 ⋅ 𝐵𝑆 4
−22.778 ⋅ 𝐵𝑆 3 + 6.3617 ⋅ 𝐵𝑆 2 + 8.3631 ⋅ 𝐵𝑆
𝑙𝑅𝑒𝑔 ( )5 ( )4
𝑡= (8) 𝑓4 = 0.2443 𝐶 𝐿
𝐶
− 0.4627 𝐶 𝐿
𝐶
𝑉 𝑀𝐺 𝑙𝑚𝑎𝑥 𝑙𝑚𝑎𝑥
( )3 ( )2 ( )
and the overall time is given as sum of the two: +0.313 𝐶
𝐶𝐿
− 0.0677 𝐶
𝐶𝐿
+ 0.0079 𝐶 𝐿
𝐶
𝑙𝑚𝑎𝑥 𝑙𝑚𝑎𝑥 𝑙𝑚𝑎𝑥

𝑡𝑇 𝑂𝑇 = 𝑡𝑈 𝑊 + 𝑡𝐷𝑊 (9)

This latter parameter is then used to identify the best Foil Section NACA 0012 (both DB and RUD)
appendage combination: the lower the time 𝑡𝑇 𝑂𝑇 , the 𝑎2𝐷 = 2𝜋 ⋅ [2 ⋅ 10−22 ⋅ 𝑅𝑒3 − 2 ⋅ 10−15 ⋅ 𝑅𝑒2
quicker will be the boat during the race. +6 ⋅ 10−8 ⋅ 𝑅𝑒 + 0.756]
It shall be remarked that the 𝑡 values so computed are 𝐶𝑙𝑚𝑎𝑥 = 2 ⋅ 10−21 ⋅ 𝑅𝑒3 − 5 ⋅ 10−14 ⋅ 𝑅𝑒2 + 3 ⋅ 10−7 ⋅ 𝑅𝑒 + 0.8473
ranking parameters and do not represent real racing 𝐶𝑑𝑚𝑖𝑛 = −8 ⋅ 10−11 ⋅ 𝑅𝑒 + 0.0065
times, since they cannot account for: i) the time needed
for tacking and gybing, and ii) the slowing action in-
duced by the rudder activation (for instance to correct Table 2
the route during upwind navigation or to bring back Daggerboards considered for the analyses.
the boat to a flat attitude in downwind conditions).
Planform 𝑑1 [m] 𝑑2 [m] 𝑑3 [m]
For a given daggerboard planform shape and draft 𝑑, the 1/4 ellipse 0.9 1.1 1.3
computation is performed on several 𝑘 ratios, until when the rectangular 0.9 - 1.3
plot 𝑡 − 𝑘 shows a minimum (see step 22 of Fig. 3). That
minimum identifies the optimum 𝑘 (i.e., the best aspect ratio
𝐴𝑅𝐷𝐵 ) for that particular draft and planform. After that, the xflr5 analyses and it was also shown by the towing tank tests
procedure is repeated for different 𝑑 and for different plan- conducted at Delft [3, 7]). On the other hand, foils with
forms; the data are then recorded and plotted together in the larger taper ratios will perform intermediately between ellip-
same 𝑡 − 𝑘 plot for comparison. The daggerboard geome- tic and rectangular ones, whereas geometries featuring low
try which shows the minimum 𝑡𝑇 𝑂𝑇 is the best (if feasible 𝜆 values (i.e., triangular or quasi-triangular wings) provide
in structural, stability and control terms - see Sec. 3) and, in poor performances [2, 10] and were not taken into consider-
turn, identifies the size of the rudder as well. ation. A dedicated paragraph is addressed to the sweep angle
and related foil performance in Sec. 5.
The results in terms of upwind, downwind and total times
3. Results and Verifications (𝑡𝑈 𝑊 , 𝑡𝐷𝑊 , and 𝑡𝑇 𝑂𝑇 ) are presented in Fig. 6 and anal-
The procedure is applied here to the case-study of a 4.6 ysed in Sec. 3.1 . The comparison between the daggerboard
m skiff, whose characteristics are reported in table 1 with performances resulting from the method and those obtained
the environmental conditions considered and the polynomial from xflr5 are then presented in Sec. 3.2, with the validation
functions used for the VPP, whereas the five daggerboard ge- against experimental data.
ometries analysed are summarized in table 2. Among the
usual planform shapes (elliptic, rectangular, and tapered), 3.1. Results
the analysis has been applied solely to the first two and only The outputs of the computations are presented in Fig. 6
to straight geometries (i.e., with zero sweep angle): in fact, for the upwind, downwind, and total times, respectively. The
tapered wings with optimum taper ratio (𝜆 ≈ 0.5) perform following observations can be reported:
similarly to the elliptic ones (this was indeed observed with

Castegnaro et al.: Preprint submitted to Elsevier Page 9 of 26


An Engingeering Method to Design Daggerboards and Rudders

1050

1000

tuw [s]
950

ellip. d=0.9 m
ellip. d=1.1 m
900
ellip. d=1.3 m
rect. d=0.9 m
rect. d=1.3 m
850
0 0.2 0.4 0.6 0.8 1
cr/d

(a) Upwind Time.


900

850
[s]

800
dw
t

ellip. d=0.9 m
ellip. d=1.1 m
750
ellip. d=1.3 m
rect. d=0.9 m
rect. d=1.3 m
700
0 0.2 0.4 0.6 0.8 1
cr/d

(b) Downwind Time.


1850

1800
ttot [s]

1750

ellip. d=0.9 m
ellip. d=1.1 m
1700
ellip. d=1.3 m
rect. d=0.9 m
rect. d=1.3 m
1650
0 0.2 0.4 0.6 0.8 1
cr/d

(c) Total Time.


Figure 6: Timing performance for a 4.6 m skiff sailing with 𝑇 𝑊 𝑆 = 5 m/s and using
elliptic and rectangular daggerboards of different draft 𝑑: the minimum in the Total time
plot identifies the best 𝑐𝑟 ∕𝑑 ratio (i.e., the aspect ratio) for a given semispan and planform
shape.

• Upwind - Independently of the planform and semis- way) to achieve the required lift, increasing both the
pan 𝑑, a minimum on the upwind leg timing is identi- viscous and inviscid resistances related to the higher
fied as corresponding to a particular 𝑐𝑟 ∕𝑑 ratio (Fig. lift coefficients involved.
6a)). This minimum is slightly appreciable for the
curves pertaining to the lowest semispan (𝑑 = 0.9 • Upwind - At equal daggerboard draft 𝑑, the elliptic ge-
m), though becoming more evident for the geometries ometries feature lower minimums than the rectangu-
at higher 𝑑. In upwind conditions (when the dagger- lar ones, although the differences are narrow in terms
board works similarly to an airplane wing) the geome- of 𝑡𝑈 𝑊 : ≈ 1.5𝑠 at 𝑑 = 0.9 m and just 0.3 s at 𝑑 =
try corresponding to this minimum represents the best 1.3 m. On the other hand, the right-hand sides of the
compromise between profile and friction losses: for a curves for the rectangular foils are steeper than the el-
given 𝑑, the geometries with higher 𝑐𝑟 ∕𝑑 ratio (right liptic ones. This is easily explained in terms of the
side of the curves) produce an excessive resistance be- higher surface (and thus, higher friction) of a rectan-
cause of the higher surface 𝑆𝐷𝐵 ; on the opposite side gular daggerboard compared to an elliptic one of equal
(left part of the curves), foils with lower 𝑐𝑟 ∕𝑑 (and draft and 𝑘 ratio. This aspect makes the identification
thus, less surface) produce less friction drag, but they of the optimum 𝑐𝑟 ∕𝑑 ratio more critical for rectangular
must operate at higher angles of attack (i.e., higher lee- foils than for elliptic ones.

Castegnaro et al.: Preprint submitted to Elsevier Page 10 of 26


An Engingeering Method to Design Daggerboards and Rudders

• Overall - Referring at Fig. 6a), increasing the draft 𝑑 are direct consequences of simple geometry: indeed,
reduces the minimum 𝑡𝑈 𝑊 , and the optimum 𝑐𝑟 ∕𝑑 ra- the surface of a rectangular foil is 27% higher with re-
tio shifts towards lower values. Looking at the elliptic spect to an elliptic geometry with the same semi-span
geometries, 𝑡𝑈 𝑊 𝑜𝑝𝑡 is reduced of 23 s (−2.5%) pass- 𝑑 [6, p.205], thus providing the same lifting capability
ing from 𝑑 = 0.9 m (𝑘 = 0.5) to 1.3 m (𝑘 = 0.25). at a lower 𝐶𝐿 (with the lower losses related) but also
On the other hand, the corresponding timing increase with higher friction drag. The effect of this increase
during the downwind leg (Fig. 6b)) due to the longer in the friction drag is well visible in the plot referring
daggerboard at 𝑘 = 0.25 with respect to the shorter to the downwind leg (Fig. 6b)), where the rectangular
one at 𝑘 = 0.5 is just 1.5 s, for a final total gain of geometries evidently require higher racing times with
21.5 s. Although such gain might appear small, ben- respect to the elliptic ones with equal 𝑑.
efits of this entity (see also [8] and [3, p.91]) would It can also be observed that the longer the dagger-
make a race chanceless for a boat using the shorter board draft the narrower becomes the range around
daggerboard. Yacht designers are aware that longer the optimum 𝑘 ratio, in particular for the rectangu-
keels provide better boat performance (e.g., [8]); how- lar foil, whereas the elliptic one shows less steep gra-
ever, these computations also allow an indirect expla- dients. As anticipated, this translates into an impor-
nation of the general rule of handicap estimation (for tant advantage when selecting an elliptical planform,
races involving yachts of different classes), which sta- since it makes the identification of the 𝑘𝑜𝑝𝑡 less critical
tistically quantifies the advantage of boats with longer with respect to rectangular ones. Although Trower [5]
appendages [29]. reports that elliptic planforms are numerous among
• Downwind - As expected, no minimum is observed in dinghies, some designers seems to be favouring dif-
the curves pertaining to the downwind leg (Fig. 6b)). ferent fashions (see Fig. 15 in Appendix).
This is reasonable, since there is no need for compen- • Overall -The relative advantage of longer geometries
sating leeway when sailing leeward (in fact, a higher 𝛾 with respect to shorter ones reduces with the increase
angle increases the 𝑉 𝑀𝐺 - see Fig. 4 in Appendix B). of 𝑑 (compare the differences between the elliptic curves
As a matter of fact, the daggerboard is generally lifted at 𝑑 = 0.9, 1.1 and 1.3, in Fig.s 6a and c). This sug-
up when sailing downwind on single-helmed crafts (as gests that an excessive elongation of the daggerboard
the Laser Radial, for instance). This same action is would end into producing negligible gains, although
not performed on skiffs, because: i) it would create it is likely that other aspects (structural, stability, or
an obstacle for the crew, ii) to dump excessive roll maximum depth allowed) will play a major role in lim-
movements during the navigation. Accordingly, on iting the maximum draft.
dinghies and skiffs with helmsman and crew the dag-
It seems fair mentioning that 𝑑 was chosen as inde-
gerboard remains in position also during the down-
pendent variable because the maximum draft is a typ-
wind leg.
ical constraint in daggerboard design, both to limit the
• Overall - When the upwind and downwind times are maximum depth (to avoid hitting the ground when ap-
summed together, the minimum is identified at lower proaching the beach/harbour) and for structural rea-
values with respect to the windward curves (compare sons. Indeed, the daggerboard behaves like a cantilevered
Fig.s 6a) and b)): for instance, for the elliptic geome- beam and too long geometries might experience ex-
tries the minimum shifts from 𝑘 = 0.5 to 0.4 with cessive stresses in case of capsizing, when the crew
𝑑 = 0.9 m, and from 0.25 to 0.2 with 𝑑 = 1.3 m (and jumps on it to bring up back the boat in vertical posi-
in analogous manner for the rectangular ones). This tion (this argument is further developed in Sec. 4).
is due to the downwind contribution, which favours
In support of the reader who is willing to re-construct the
smaller daggerboards to reduce the drag during the
VPP algorithm using the equations reported in Appendix B,
leeward navigation. In spite of this effect, the lower
the numerical results pertaining to the rectangular dagger-
impact of the downwind leg on the overall time with
board (𝑑 = 1.3 m and 𝑘 = 0.4) derived from the first, sec-
respect to the upwind one (compare Fig. 6a) and b))
ond, and final computations are reported in table 9.
provides an explanation of the common custom of re-
The method can certainly be used considering different en-
ferring the design of the appendages exclusively to the
vironmental conditions: with this regard, it was applied to
windward conditions.
the same 4.6 skiff but with TWS = 4 and 6 m/s, respectively,
• Overall - Regarding the planform shape, by compar- and different wave systems: the results are reported in Fig.
ing the upwind and, most importantly, the overall per- 17 in Appendix B for 𝑑 = 1.3 m rectangular and elliptic
formance (Fig. 6a) and c)) of rectangular and elliptic foils, showing analogous trends with respect to those of Fig.
geometries, it can be observed that the minimum tim- 6. However, it is here remarked that being aware of the re-
ing values are approximately equals for foils with the lation between wind and sea conditions (i.e., the wave sys-
same draft. However, for rectangular daggerboards tem created by a certain wind speed) is fundamental to ob-
the 𝑘𝑜𝑝𝑡 occurs at lower values with respect to the ellip- tain representative indications for the targeted race field. At
tic ones and the curves are steeper at higher 𝑐𝑟 ∕𝑑. Both last, although applied here to a windward-leeward race, the

Castegnaro et al.: Preprint submitted to Elsevier Page 11 of 26


An Engingeering Method to Design Daggerboards and Rudders

method could be also applied to different routes if a different


regatta scheme is involved. In that case, the computation will
include additional legs (abeam, for instance) and the identi- 0.8
fication of the optimum geometry will take into account the
related times, as well. 0.6

CL [−]
3.2. Comparison with xflr5 for the Upwind data. 0.4
AR 5.14, xflr5
As first verification, it was deemed important verifying AR 5.14, exper.
0.2
the VPP polar prediction capability for the foils at different AR 10.07, xflr5
Reynolds. Accordingly, the daggerboard upwind results per- 0
AR 10.07, exper.
0 5 10 15
taining to different 𝑘 ratios have been compared with data α [°]
obtained from xflr5 analyses on analogous geometries. The
comparison has been performed on upwind cases only, being
the downwind ones more trivial due to the higher boat speeds (a)
(i.e., higher Re) and the lower angles of attack involved. At
first, the polar prediction capability of xflr5 has been preven-
tively verified against experimental data; this was retained 0.06 AR 5.14, xflr5
necessary because of the relatively low Reynolds numbers AR 5.14, exper.
typical of small yachts appendages. AR 10.07, xflr5
0.04 AR 10.07, exper.

CD
3.2.1. xflr5 validation
xflr5 is an open-source program for the design and anal- 0.02
ysis of sailplane models based on the well-known Xfoil soft-
ware [20]: this is a panel code that integrates a solver of the 0
Euler equations (for the inviscid flow) with a two-equations 0 0.2 0.4
CL
0.6 0.8

integral boundary layer formulation and a transition model


[30]. A detailed description of both Xfoil and xflr5 can be
found in the related literature [16, 20, 30]. (b)
It was deemed important assessing the polar prediction capa-
bility of xflr5 at low Re and the possible tuning of some sim-
ulation parameters. However, experimental data of wings
within the range 5 ⋅ 105 < 𝑅𝑒 < 3 ⋅ 106 are not numerous in 30
the available literature; with this regard, we referred to the
tests presented in the NACA report [31] on two backswept
CL/CD

20
geometries, the characteristics of which are reported in table AR 5.14, xflr5
3, together with the setting-up parameters of the xflr5 anal- 10 AR 5.14, exper.
yses. AR 10.07, xflr5
AR 10.07, exper.
0
0 5 10 15
Table 3 α [°]
The geometrical characteristics of the two wings and the model
parameters used for the xflr validation.
(c)
wing 1 wing 2
section 65A012
Λ 35° Figure 7: Comparison of xflr5 data with respect to experiments
𝑐𝑟 412 mm (see [31]) on two sweptback wings (Λ = 35°) tested at 𝑅𝑒 =
AR 10.07 5.14 2 ⋅ 106 and Ma=0.25. The wings feature 𝐴𝑅 ≈ 5 and ≈ 10 and
𝜆 0.500 0.713 the NACA 65A012 section.

xflr5 analysis parameters


𝑅𝑒 2 ⋅ 106 ers) in terms of 𝐶𝐿 − 𝛼, 𝐶𝐷 − 𝐶𝐿 , and 𝐶𝐿 ∕𝐶𝐷 − 𝛼. In gen-
𝑀𝑎 0.25 eral, a better agreement between simulations and experimen-
𝑁𝑐𝑟𝑖𝑡 9 tal data is observable for the wing with AR=5 with respect to
type of analysis Vortex lattice Method/Panel - viscous the AR=10 one; moreover, a satisfactory match is visible for
wing nr. of panels 25 both geometries at low angles of attack (< 2°), but with an
increasing divergence at higher angles. In particular, it can
be observed (see Fig. 7a) that the slope of the experimental
In Fig. 7, the results of the xflr5 analyses (solid lines) are lift coefficient curves decreases with increasing values of the
compared with the experimental data (dotted lines with mark-

Castegnaro et al.: Preprint submitted to Elsevier Page 12 of 26


An Engingeering Method to Design Daggerboards and Rudders

d [m] 𝐶𝐿 BS [m/s] 𝑐𝑟 ∕𝑑 𝑅𝑒 ⋅ 106 𝐶𝐷 𝛾 [°] 𝐶𝐷𝑥𝑓 𝑙𝑟5 𝛾𝑥𝑓 𝑙𝑟5 [°] Δ𝛾 (%) Δ𝐶𝐷 (%)
Rectangular 0.9 0.335 3.405 0.40 1.405 0.015 5.1 0.013 4.7 8% 11%
1.3 0.327 3.402 0.20 1.014 0.011 4.3 0.010 3.7 14% 12%
Elliptic 0.9 0.335 3.405 0.50 1.405 0.013 4.7 0.013 4.7 7% 10%
1.3 0.332 3.407 0.25 0.963 0.011 4.3 0.010 3.7 14% 12%
Table 4
Comparison of Computed data versus xflr5 simulations at the 𝑘 ratios which provide the
minimum upwind times.

angle of attack, diverging from the linear behaviour shown ical data pertaining to this latter reported in table 4. The
by the numerical analyses. This behaviour is more evident computed trends are well in agreement with the results of
for the wing with AR=10 with respect to the AR=5 one, the simulations, though showing some differences in terms
and is ascribable primarily to the stall of the outer portions of values: in particular, the computed results overestimate
of the wings typical of geometries with relevant sweep-back the leeway angles with respect to xflr5, and this translates
angles, and, to a minor extent, to small distortions of the also into higher drag coefficients 𝐶𝐷 . Focusing on Fig.s 8 a)
wing models because of torsion and bending during the tests. and b), it can be observed that both the elliptic and rectan-
Both phenomena are typical of highly sweptback wings [31] gular geometries show the largest difference Δ𝛾 at the low-
and they cannot be predicted by the xflr5 code. Nonethe- est 𝑘 ratio, but with a constant reduction with increasing 𝑘
less, this is of little importance within the scope of this work, values. Furthermore, the differences are higher for the foils
as racing daggerboards normally neither feature such large with d=0.9 m than the 1.3 m ones: being the aspect ratios
sweep-back angles nor they are usually designed to work at equals for the longer and shorter daggerboards (in fact, 𝐴𝑅
angles of attack (leeway) ⪆ 6°. Focusing on the comparison is function of 𝑘 - see the Appendix B), this behaviour of
at lift coefficients 𝐶𝐿 < 0.4, a good matching is observ- Δ𝛾 cannot be explained but in terms of 𝑅𝑒. As previously
able between numerical and experimental data in terms of mentioned, indeed the Reynolds number is proportional both
𝐶𝐿 − 𝐶𝐷 curves, thus indicating a good prediction also of to 𝑘 and 𝑑, ranging from ≈ 3.5 ⋅ 105 at the lower 𝑘 up to
the viscous drag. On the other hand, a certain disagreement ≈ 1.4 ⋅ 106 for 𝑘 = 0.5 for the d=0.9 m foils. Therefore, the
is observable at the edge of the laminar bucket (𝐶𝐿 ≈ 0.23), differences Δ𝛾 between computation and simulations can be
where xflr5 overestimates the 𝐶𝐷 of the wing with AR=10 explained by the approximation introduced by the polyno-
(+35%), while it slightly underestimates (-13%) the same pa- mial (table 1) used to fit the experimental data describing
rameter for the wing with AR=5. This is the reason of the the trend 𝑎2𝐷 = 𝑓1 (𝑅𝑒) for the 0012 section, visible in Fig.
marked difference between the peak values of the 𝐶𝐿 ∕𝐶𝐷 5a. It shall also be remarked that the foil section data on
curves (Fig. 7c) for the wing with AR = 10, while those which the polynomial of table 1 was built are limited to a
with AR=5 show an acceptable agreement. In general, this minimum 𝑅𝑒 = 7.0 ⋅ 105 (see Fig. 5), thus the performances
is still considered a fair agreement between experiments and below this value are extrapolated and cannot be considered
xflr5 simulations. However, to avoid the uncertainty related reliable. On the other hand, neither xflr5 has been validated
to laminar foils operating at the edges of the low-drag buck- at such low Reynolds numbers, therefore a certain degree of
ets, it was decided to use a conventional NACA 0012 section uncertainty relies at those regimes. In spite of this, at higher
within this work. Reynolds the agreement between the results of the compu-
Analyses on the two wings were repeated also at 𝑅𝑒 = 4⋅106 tations and simulations considerably improves, with differ-
providing similar comparisons, whereas simulations with lower ences Δ𝛾 < 14% and Δ𝐶𝐷 < 11% (see table 4) around the
𝑁𝑐𝑟𝑖𝑡 < 9 showed worse matching between the experimental optimum 𝑘 ratios.
and numerical data. Accordingly, the prediction capability As the differences between the computations and the re-
of the xflr5 code is considered validated within the objec- sults from xflr5 are higher at lower k ratios, the VPP is ex-
tives of this work. pected to excessively penalize the geometries with low 𝑐𝑟 ∕𝑑
(thus, with higher aspect ratios), especially if the Reynolds
3.2.2. Comparison between the computation results number falls below 7⋅105 . As a consequence, in that case the
and xflr5 data on the same geometries 𝑘𝑜𝑝𝑡 might result slightly displaced towards the right side of
Elliptic and rectangular foils with drafts equal to 0.9 m the curves. Nonetheless, the comparison reported in Fig. 9
and 1.3 m, featuring analogous k ratios of those in the com- in terms of 𝑡𝑈 𝑊 computed using xflr5 and the performance
putations have been reconstructed into xflr5. The analyses parameters from the VPP for the rectangular foil with d =
were performed imposing the same boat speed and lift re- 1.3 m and 𝑘 ratios around the optimum one shows that the
sulting from the computations (thus, at equal lift coefficients trends are fairly in agreement, with a maximum difference of
and Reynolds numbers - see table 4); by doing so, the data 9.2. s (≈ 1%) at 𝑘 = 0.15. Differences of this magnitude are
are compared in terms of angles of attack (leeway angles) acceptable within an engineering approach as the one pro-
and drag coefficients. The results are plotted in Fig. 8 for posed in this work, aimed at the first sizing of the appendage.
different 𝑘 ratios around the optimum one, with the numer- If necessary, further analyses can be later performed with

Castegnaro et al.: Preprint submitted to Elsevier Page 13 of 26


An Engingeering Method to Design Daggerboards and Rudders

(a) (b)

(c) (d)
Figure 8: Comparison between the daggerboard performance in terms of leeway angle 𝛾
and drag coefficient 𝐶𝐷 ) computed by the VPP (solid lines) and the same parameters
obtained with xflr5 (markers and dotted lines) on the same foils.

4. Rudder Design and Final Verifications


930
When the iterations are completed and an optimum dag-
gerboard geometry has been identified, the rudder surface
910
remains defined as well, and its design can be finalized ac-
tuw [s]

cording to the indications reported in this paragraph. At last,


890
VPP some structural and stability verifications, reported in the
870
xflr5 second part of the paragraph, shall be done before conclud-
0 0.1 0.2 0.3
cr/d
0.4 0.5 0.6 ing the design exercise.

Figure 9: Comparison between upwind times at different 𝑐𝑟 ∕𝑑


4.1. Indications on Rudder Design
ratios computed using xflr5 (in blue) or the VPP (in red) for As anticipated, the rudder design shall respond to two
the rectangular daggerboard with d=1.3 m. conflicting requirements: low drag during navigation, as well
as delaying as much as possible the stall to ensure a high
range of maneuverability (also at low speeds), with this lat-
more complex tools (e.g., by RANS simulations) to identify ter being the preeminent one [5, 7, 8]. Furthermore, un-
with higher accuracy the optimum geometry. like the daggerboard which is expected to work at rather
constant conditions, the rudder is continuously activated to
keep the route. This dynamic operation further complicates
a possible design based on a performance estimation simi-
lar to the daggerboard one. Although still in principle pos-

Castegnaro et al.: Preprint submitted to Elsevier Page 14 of 26


An Engingeering Method to Design Daggerboards and Rudders

sible, it would require a law for the expected activation of


the tiller during the race (actually, several laws depending
on the weather conditions, to allow the identification of the
best trade-off one). Such a law database is not available and
its assumption would risk to be erroneous; in light of this,
we provide here information which allows the definition of
the rudder design, on one side, and might help developing a
performance estimation method also for the rear foil, on the
other.
There is a general agreement among the authors that the rud-
der shall be designed primarily to delay the stall. To account
also for an efficient navigation, we suggest selecting the rud-
der foil section comparing the parameter 𝐶𝑙𝑚𝑎𝑥 ∕𝐶𝑑𝑚𝑖𝑛 for the
different candidates (the higher, the better). Furthermore,
the following aspects shall be taken into account:
• Being the rudder hung on the transom, it crosses the
water surface; therefore, unlike the daggerboard, there
is not a solid surface acting like a symmetry plane (i.e.,
no mirroring concept). Accordingly, the aspect ratio Figure 10: Different chord reductions in the rudder re-
𝐴𝑅𝑅 of the rudder during navigation shall not be dou- gion close to the water surface: the rudder of a 470
bled as the daggerboard one (this aspect complicates (on the left) and a design by the authors (on the right)
also the numerical simulations on the appendage com- during a towing tank test.
binations, since only one of the two foils can be mir-
rored, introducing the necessity of particular precau-
tions on the domain definition and on the boundary distance 𝑥 lower in the case of the front foil. However,
conditions). Furthermore, this simultaneous opera- some daggerboard and keel designs feature as well a
tion of the rudder between two fluids causes a problem chord reduction close to the root [5], whereas in this
known as ventilation, with the water level lowering on respect Vacanti [8] suggests a sort of transition fairing
the suction side of the foil during steering (because of (although this feature seems rather difficult to apply
the lower local pressure), with a consequent reduction on dinghies). Qualitatively, the low velocity region of
of the effective rudder surface. Some designers pro- the hull boundary layer can be observed when some
pose the use of fences to partly compensate this effect air bubbles slowly slip along the hull surface during
[5], although such devices cannot cope with the fol- navigation (see Fig. 11).
lowing issue.
• As already reported, the rudder is placed in line be-
• Being placed at the very end of the craft, the first un- hind the daggerboard, thus facing dirty flow condi-
derwater sections of the rudder are immersed into the tions characterised by the wake and the downwash of
hull boundary layer. This low velocity region in the the first foil. Researchers from Delft [25] proposed an
fluid around the foil makes part of the rudder less ef- equation to estimate the downwash angle due to the
fective in the generation of lift. The flatness of the hull daggerboard as:
surface of most dinghies and skiffs allows, in first ap- √
proximation, to estimate the vertical extension of this 𝐶𝐿𝐷𝐵
region 𝛿𝐵𝐿 from the equation for the turbulent incom- Φ𝐷𝐵 = 0.192 (11)
2 ⋅ 𝐴𝑅𝐷𝐵
pressible flow over a flat plate (see [22]):
Therefore, the angle of attack seen by the rudder could
0.37 ⋅ 𝑥
𝛿𝐵𝐿 = (10) be estimated by:
1∕5
𝑅𝑒𝑥
𝛼𝑅 = 𝛾 − Φ𝐷𝐵 (12)
where 𝑅𝑒𝑥 for the rudder is the Reynolds number com-
puted on 𝑥 = 𝐿𝑊 𝐿 . Depending on the boat speed 𝐵𝑆, The angle of attack reduction identified by Eq. 12 is
the vertical extension of this low-velocity region can one of the causes of the lift decrease Δ𝐿𝑅 observed
affect a not-negligible part of the rudder draft (up to in [26] by comparing the rudder in isolated operation
≈ 8−10 cm). To address both this issue and the venti- and in line with the daggerboard.
lation one, we suggest a reduction of the chord lengths
in the upper part of the rudder (see Fig. 10), as much • Unlike the daggerboard, the maximum draft is not a
as permitted by structural integrity. Also the dagger- major constrain for the rudder, which in any case is
board root is immersed into the hull boundary layer, normally shorter and is not used to flip back the boat
though for a reduced vertical extension 𝛿𝐵𝐿 , being the after a capsize. High aspect ratios are more efficient

Castegnaro et al.: Preprint submitted to Elsevier Page 15 of 26


An Engingeering Method to Design Daggerboards and Rudders

where 𝐼 [𝑚𝑚4 ] is the moment of inertia of the foil sec-


tion and 𝑦𝑚𝑎𝑥 [𝑚𝑚] is the distance between the dagger-
board axis (along the centroids of the sections) and the
point of maximum thickness on the root cross section.
Assuming a crew weight 𝑊 = 80 − 90 [kg] standing
on the daggerboard tip normally makes unnecessary
repeating the verification also for the distributed lift-
ing load during navigation.
ii) Since dinghies and skiffs do not have a bulb that pro-
vides ballast, the crew and the helmsman shall en-
sure the necessary righting moment moving their own
Figure 11: Air bubbles sliding along the hull can provide a weight on the deck and, in case, using the trapezes.
qualitative indication of the boundary layer. In theory, an excessively long daggerboard might shift
the centre of hydrodynamic efforts so deeply to induce
an excessive heeling moment 𝐻𝑀 with respect to the
but are more prone to stall, also because of the lower maximum achievable righting moment 𝑅𝑀𝑚𝑎𝑥 (see
Reynolds numbers related to the shorter chord lengths Fig. 12a) . Accordingly, it shall be verified that at the
at equal rudder surface 𝑆𝑅 . maximum boat speed expected:
• Without limitations on the maximum draft, the ellip-
tic planform is expected to be the most efficient one 𝑅𝑀𝑚𝑎𝑥 > 𝐿𝐷𝐵 ⋅ 𝑧𝐷𝐵 + 𝐹𝑠 ⋅ 𝑧𝑆 (14)
for the rudder, and indeed is largely employed [6].
where 𝐿𝐷𝐵 is the daggerboard lift, while 𝑧𝐷𝐵 and
However, it is reported [5, p.111] that squaring-off the
𝑧𝑆 are the vertical positions of the hydrodynamic and
rudder planform gives some warnings of an incipient
aerodynamic-sail centres (see [5] for practical ways to
stall to the helmsman. This is in agreement with the
identify these positions) with respect to an arbitrary
different stall mechanisms of rectangular wings with
plane of reference (the waterline, for instance). Since
respect to elliptic ones [10, p.315], where the stall is
𝐿𝐷𝐵 = 𝐹𝑠 , Eq. 14 becomes:
rather uniform along the entire span instead.
Once the rudder surface 𝑆𝑅 is obtained from the computa- 𝑅𝑀𝑚𝑎𝑥 > 𝐿𝐷𝐵 ⋅ (𝑧𝐷𝐵 + 𝑧𝑆 ) (15)
tion of Sec. 2 and accounting for the indications above, the
design finalization ultimately requires the identification of The maximum righting moment 𝑅𝑀𝑚𝑎𝑥 depends on
the aspect ratio 𝐴𝑅𝑅 representing the best compromise be- the characteristics of each boat (the maximum width,
tween stall-delaying and navigation efficiency. In lack of a the inclusion of lateral wings, the use of trapezes, etc.),
mathematical method to identify this compromise, we sug- thus it is here left in a generic form, with its derivation
gest referring to the benchmarking in appendix A for the being trivial given the boat specifications.
selection of a range of first guess, with the related advan- iii) The boat balance along the yaw has to be targeted dur-
tage/drawbacks to be compared one by one (for instance by ing the upwind conditions corresponding to the ex-
numerical analyses with xflr5 on the isolated rudder geome- pected 𝑇 𝑊 𝑆, since the beam wind and leeward courses
try). are less demanding to control the craft. We follow
here the approach proposed by Trower [5, p.120] who
4.2. Final Verifications
suggests that a good balance for a dinghy is achieved
The identified daggerboard geometry shall comply with
placing the sail overall centroid a small distance ahead
three final requirements: i) structurally, it shall withstand the
(5% of the 𝐿𝑊 𝐿 - also named lead) of the underwater
maximum expected stresses; ii) the maximum heeling mo-
centroid, this latter being identified omitting the rud-
ment 𝐻𝑀 must be lower than the highest achievable right-
der (Fig. 12b). This is a customary approach and other
ing moment 𝑅𝑀; and iii) the daggerboard position shall al-
designers prefer including the rudder as well, although
low the balance along the yaw angle (boat centering).
this would change the amount of lead recommended
i) Normally the most critical loading condition for the to achieve the proper balance.
daggerboard is in case of capsize, when the crew jumps
on the daggerboard to flip the boat back to the vertical It shall be remarked that the optimum design deriving from
position. Since the crew stands on the foil tip (to in- the VPP computation might not comply with the previous
crease as much as possible 𝑅𝑀), the maximum bend- verifications: for instance, depending on the materials and
ing moment is 𝐵𝑀 = 𝑊 ⋅ 𝑔 ⋅ 𝑑 [𝑁𝑚𝑚]. Thus, the ad- manufacturing methods employed, the maximum stresses might
missible stress 𝜎𝑎𝑑𝑚 of the daggerboard structure shall exceed 𝜎𝑎𝑑𝑚 if a foil section with insufficient thickness (e.g.,
be (with a reasonable safety margin): 6% of the chord length) is used at the at the daggerboard root.
Accordingly, these final checks are always needed before fi-
𝐵𝑀 nalizing the design task.
𝜎𝑎𝑑𝑚 > ⋅ 𝑦𝑚𝑎𝑥 [𝑀𝑃 𝑎] (13)
𝐼
Castegnaro et al.: Preprint submitted to Elsevier Page 16 of 26
An Engingeering Method to Design Daggerboards and Rudders

Table 5
Suggested criteria for a preliminary identification of the foil
sections for daggerboard and rudder.
suggested criteria
𝐶𝑙 𝑡𝑈 𝑊 1 𝑡𝐷𝑊
FSs
Daggerboard ⋅ + ⋅
𝐶𝑑 𝑡𝑇 𝑂𝑇 𝐶𝑑𝑚𝑖𝑛 𝑡𝑇 𝑂𝑇
𝐶𝑙𝑚𝑎𝑥
Rudder
𝐶𝑑𝑚𝑖𝑛

the rudder are suggested in the literature (see for instance


2W*g
LDB
[3, 5, 7, 8]). A deep analysis of the characteristics related
to the different sections is rather complex and beyond the
(a) (b) scope of this work. However, seeking for quantitative cri-
teria to identify the best candidates, the formulas of table 5
Figure 12: The equlibirium ensured by the righting moment
are tempted. Some iterations are needed for the daggerboard
RM (on the left). On the right, the Boat Centering scheme
according to Trower [5]: the sail centroid shall be placed 5% of
criterion, as a first run of the computation with an arbitrary
𝐿𝑊 𝐿 ahead of the hydrodynamic center (omitting the rudder). selected section shall be performed to obtain a first estima-
𝐶
tion of 𝑡𝑈 𝑊 , 𝑡𝐷𝑊 , and 𝑡𝑇 𝑂𝑇 . It shall be remarked that 𝐶 𝑙 in
𝑑
table 5 shall be taken at the angle of attack 𝛼0 seen by the
5. Discussions foil section (i.e., the angle of attack of the wing reduced by
the effect of the induced velocity - see [1, p.132]) resulting
In spite of the satisfactory indications provided by the from the computation. The application of the criteria of ta-
comparison with xflr5, this cannot be considered a validation ble 5 to a range of candidates allows the identification of the
of the entire method. Indeed, a definitive verification would section which maximises the parameters, as shown in Fig.
not consider the daggerboard performance alone, rather it 13 for the NACA 4-digit family with different thicknesses.
would assess the behaviour of the entire system (sailboat In particular, Fig. 13a shows that thinner sections should be
+ appendages) in terms of boat speed and leeway angle (to preferred for the daggerboard (if feasible from a structural
compute the VMG), systematically varying the daggerboard point of view and if the design angle of attack presents a
draft, 𝑐𝑟 ∕𝑑 ratio, and planform shape for given wind and ma- reasonable margin from stall). On the other hand, Fig. 13b
rine conditions. Although in principle possible, such vali- shows that the 12% thick foil section provides the best trade-
dation requires demanding means (as a towing tank) which off between maximum lift capability and minimum drag for
are beyond authors’ possibilities; consequently, this design the rudder.
method cannot be deemed entirely verified. We are aware As previously reported, some authors ([5, 7, 8]) suggest that
that this might be considered a weak point of the paper; how- the stall delaying capability should be the major design driver
ever, the structure, the physical concepts and models im- for the rudder: accordingly, sections with relatively thin lead-
plemented in the procedure have a general validity indepen- ing edges (as the NACA 6-series, for instance) should be
dently of the experimental validation, which could be done discarded in favour of families with a thicker front (as the
on different small sailboats than the one considered here. NACA 4-digit family). With this regard, the higher 𝐶𝑙𝑚𝑎𝑥
The authors would appreciate and technically support who- capability of the 0012 section is evident when compared to
ever, with such possibilities, might be interested into ad- the 64012 one at any Reynolds number (see Fig. 5c).
dressing such work of validation. Of course, all the parameters included in table 5 change
Some further comments on specific design aspects such according to Re. However, preliminary indications suggest
as the foil section identification and the sweep angle are lastly that the identification of the foil sections according to these
reported. The reader will have likely noted that the perfor- criteria is affected to a minor extent by 𝑅𝑒 within the typical
mance evaluation procedure relies on assuming the foil sec- ranges of sailing foils (i.e., the section maximising the for-
tion for both the daggerboard and the rudder. Different sec- mulas in table 5 is the same at different Reynolds numbers).
tions can be compared by changing the polynomials which Further investigations are needed, however, and the applica-
describes the airfoil performance with 𝑅𝑒 and repeating the tion of the criteria at the extremes of the expected range of
computations. Nonetheless, it is deemed important ranking 𝑅𝑒 is recommended for verification.
the best candidates in a paragraph dedicated to the topic. The surface finishing shall be as smooth as possible, al-
though polishing or waxing is useless in terms of perfor-
5.1. Foil section mance differences with respect to well sanded surfaces, ac-
The identification of the appropriate foil sections is a dif- cording to [32]. In table 6 measurements of surface rough-
ficult design decision among the numerous options available ness along the chord direction for two daggerboards manu-
[5]: several families that can be used for the daggerboard and

Castegnaro et al.: Preprint submitted to Elsevier Page 17 of 26


An Engingeering Method to Design Daggerboards and Rudders

110 Sec. 2 work at leeway angles 𝛾 ≲ 3°. Otherwise, the drag

tT OT
tDW
100 NACA 0006 coefficient 𝐶𝑑 of laminar sections would be underestimated
·
Cdmin
90 for operations outside the low-drag bucket (but the use of
1

laminar geometries loses of sense in this latter case).


NACA 0009

80
+

NACA 0012

70
tT OT
tU W

NACA 0015 5.2. Sweep Angle


60 NACA 0018
Although the procedure allows to include the effect of
Cd
Cl
·

the sweep angle Λ into the daggerboard performance (see


NACA 0021
50
0 1 2 3 4 5 6 7
Equations 24 in Appendix B), the analysis has been limited
(a) Daggerboard. so far only to foils with straight planforms. It is nonetheless
200 important addressing a specific paragraph to the vexed ar-
180
NACA 0012
gument of the effects of the sweep on foil performance, in
light of the fact that several daggerboards and rudders fea-
NACA 0009

160 NACA 0015


ture backswept geometries, both for rectangular, elliptic and
Cdmin
Clmax

140 NACA 0006 NACA 0018 tapered planforms (see Fig. 15). Among the authors the
120
indications are rather in contrast: Larsson and Eliasson [3,
NACA 0021
p.109] and Marchaj [10, p.348] indicate that the sweep an-
100
0 1 2 3 4 5 6 7 gle is used to obtain nearly elliptic force distributions (and
thus, induced drag) with tapered foils. These authors in-
(b) Rudder.
dicate a relation Λ = 𝑓 (𝜆) to identify the optimum sweep
Figure 13: Application of the criteria of table 5 to some angle for a given taper ratio: in particular, for taper ratios
sections from the 4-digit NACA family. The x axis indicate 𝜆 ≈ 0.45 − 0.5 (which is the tapered geometry with closer
the airfoil database numbering. Airfoil data from [15] at performance to the elliptic one, see [3, p.110]) the sweep an-
𝑅𝑒 = 3.2 ⋅ 106 . gle is zero. This appears in agreement with the indications
from Brewer [33], who suggests sweepback angles reducing
Table 6 down to nearly zero with the increase of the aspect ratio 𝐴𝑅.
The surface roughness (along the chord direction) parameters Of the opposite vision appears Vacanti [8], who claims that
of two daggerboards built for the 1001Vela race. Λ is the most important parameter to increase the upwind
gel coat painted characteristics of a keel, and provides the following equa-
tion (reported also in [34]) to compute the lift curve slope of
Ra [𝜇m] 0.46 0.14
a swept foil:
dev. st. [𝜇m] 0.14 0.05
Rz [𝜇m] 3.01 0.85 𝐶𝐿 2𝜋 ⋅ 𝐴𝑅
= √( (16)
dev. st. [𝜇m] 0.74 0.14 𝛼 )2
𝐴𝑅
2+ 𝑘
⋅ [1 + (𝑡𝑎𝑛Λ50 )2 ] + 4
𝑣

where Λ50 is the sweep angle of the positions at mid-chord,


factured for the 1001Vela competition and treated with dif-
𝑘𝑣 is a scale factor1 , and 𝑡𝑎𝑛Λ50 = 𝑡𝑎𝑛Λ − 𝐴𝑅 4 1−𝜆
⋅ 2(1+𝜆) .
ferent coatings are reported. The data clearly show the su-
periority in terms of surface finishing of the painting with Equation 16 resembles the form of Eq. 24 by Kuchemann
respect to the sole gel coat. [23] (reported in Appendix B) for rectangular foils (where
A drawback of the method is that the airfoil polars at 𝑎2𝐷 = 2𝜋, in the thin airfoil theory [22])2 . Unlike Eq. 24
several Reynolds are required and these are available from however, Equation 16 identifies the highest lift curve slope
experimental data for a limited number of sections (see e.g., 𝐶𝐿 ∕𝛼 at sweep angles Λ higher than zero for tapered wings
[15, 18, 19]). In lack of these, the section polars can be ob- (see Fig. 14), and generally around 15° according to Va-
tained from xflr5 (with the possibility of integrating an au- canti. Also Marchaj, although admitting that the inclusion of
tomatic polar prediction in the iterative loop of Fig. 3, as swept keels/daggerboards might be due to aesthetic reasons
done in [24] for instance). Particular care shall be taken how- rather than to increase the foil efficiency [10, p.340], in a
ever, especially with respect to 𝐶𝑙𝑚𝑎𝑥 estimated values, and a caption at p. 349 claims that for moderately tapered geome-
preliminary validation against experimental data on similar tries the highest 𝐶𝐿 ∕𝛼 occurs with sweptback foils rather
sections is always recommended. than straight ones. On the other hand, the same author also
At last, it shall be notified that the generalized polar con- 1 Vacanti [8] does not provide an explanation of the scale factor 𝑘
𝑣
cept developed by Jacobs & Sherman [15] is precedent to which divides the aspect ratio 𝐴𝑅 at the denominator of Eq. 16, though
the development of laminar foils, thus it cannot account for reporting a value rather close to one (1.003). In [34] the same factor is
𝐶 ∕𝛼
the sudden performance variation occurring outside the edge defined as 𝑘𝑣 = √𝐿
2
, where 𝑀𝑎 is the Mach number, although
2𝜋∕ (1−𝑀𝑎 )
of the low-drag bucket. This should not represent an issue claiming that some authors use 𝑘𝑣 = 1.
2 Both Eq.s 24 and 16 assume a linear lift curve slope, thus not taking
for laminar foils with a sufficient bucket extension (≈ ±3°),
into account the non-linear behaviour at larger angles of attack shown by
if the optimum planforms identified by the computations of highly swept wings as those tested in [31].

Castegnaro et al.: Preprint submitted to Elsevier Page 18 of 26


An Engingeering Method to Design Daggerboards and Rudders

swept foils (of any planform shape) with respect to straight


5
ones, for the typical aspect ratios pertaining to the dagger-
boards of racing dinghies and skiffs. As an indirect confir-
4 mation, all the foils of the 1001Vela Race cup prototypes are
CL/α [1/rad]

straight ones, according to the authors’ best knowledge. On


3 the other hand, the sweep angle might represent a potential
design solution to provide an acceptable rudder behavior be-
2 Kuchemann AR=6.67 yond the stall (see [10, p.347]), as well as to shift longitudi-
Vacanti AR=5.14 λ = 0.713
Vacanti AR=10.07
nally the hydrodynamic center of the underwater actions, in
1 Rect. wing AR=6.67 xflr5 case of constraints on the positions of the daggerboard and
Taper. wing AR=5.14 λ = 0.713 xflr5
Taper. wing AR=10.07 λ = 0.500 xflr5 rudder.
0
−20 −10 0 10 20 30 40 50 60
Λ [°]
6. Conclusions
Figure 14: Lift curve slopes for foils with different sweep angles
(simulated with xflr5) compared with the trends provided by
An engineering method, based on the fundamental equa-
Eq.s 24 and Eq. 16. The rectangular foils were simulated at tions of the fluid-dynamics of sailing, has been developed to
𝑅𝑒 = 1𝑀 (𝑎2𝐷 = 5.73) and the two tapered ones at 𝑅𝑒 = 2𝑀 provide in a rather simple way a unified comparison of the
(𝑎2𝐷 = 6.07). The foil section is the NACA 65A012 for all the basic design choices (i.e., basic planform shapes, drafts, and
geometries. aspect ratios) for the appendages of small racing yachts. The
method is based on a Velocity Prediction Program which es-
timates the sum of both the upwind and downwind times dur-
presents experimental plots of 𝐶𝐿 versus 𝛼 for three rectan- ing the race, accounting also for the effect of the Reynolds
gular geometries with 𝐴𝑅 = 6 and Λ = to 0°, -20°, and number variation of the foil performance, and it can be eas-
+20°: the geometry featuring the highest 𝐶𝐿 ∕𝛼 is still the ily implemented without requiring complex numerical tools.
straight one, in agreement with the indications provided by For a given daggerboard shape and draft, the procedure pro-
Eq. 24. vides the optimum aspect ratio by identifying the root chord-
In this somehow confusing scenario, we compared the to-draft ration 𝑐𝑟 ∕𝑑 which minimizes the sum of the upwind
trends resulting from Eq.s 16 and 24 with the lift curve slopes and downwind times. The method has been applied to the
of tapered and rectangular foils with different sweep angles case of a 4.6 m skiff sailing with a true wind speed of 5 m/s,
as obtained from xflr5. In particular, we analysed the cases showing that:
of rectangular foils with AR=6.67 and, again, the two ta- • An optimum trade-off exists between upwind and down-
pered geometries with AR = 5.14, 𝜆 = 0.713 and AR = wind conditions, identified in terms of the optimum
10.07, 𝜆 = 0.500 from the NACA report [31]. The results 𝑐𝑟 ∕𝑑 ratio which minimizes the overall racing time.
are reported in Fig. 14: the trend resulting from Eq. 24 is
well in agreement with the numerical data, showing a sym- • In general, the longer the daggerboard, the faster the
metric behaviour with respect to the zero sweep angle, where boat: in the elliptic case analysed the overall racing
the rectangular foil achieves the highest 𝐶𝐿 ∕𝛼 value. On the time is reduced of 21.5 s by increasing the draft 𝑑 from
contrary, the trends resulting from Eq. 16 do not match with 0.9 to 1.3 m. The corresponding optimum 𝑐𝑟 ∕𝑑 ratio
the xflr5 data pertaining to the tapered wings; in particular: decreases from 0.5 to 0.25.
i) the peak of the curves is shifted towards higher sweep an-
• The downwind contribution shifts towards lower val-
gles (at 21° and 11°, respectively, for the wing with AR=5
ues the optimum 𝑐𝑟 ∕𝑑 identified considering only the
and the one with AR=10), whereas the results from the sim-
upwind leg (from 0.25 to 0.2 for the elliptic foil with
ulations on both tapered wings again show that the peaks
𝑑 = 1.3 m and from 0.5 to 0.4 for 𝑑 = 0.9 m). On
occur at Λ = 0°, and ii) the maximum value provided by
the other hand, the lower downwind contribution to
Eq. 16 is overestimated (of ≈ 8% and ≈ 5%), due to the use
the overall time explains why appendages are usually
of 2𝜋 at the numerator instead of the actual lift curve slope
designed only considering windward conditions.
𝑎2𝐷 corresponding to that particular Reynolds. In light of
these indications, Eq. 16 appears incorrect to estimate the • Daggerboards of the same draft but with elliptic and
𝐶𝐿 ∕𝛼, at least when used for the incompressible conditions rectangular planforms present similar overall timing
(𝑀𝑎 ≈ 0) pertaining to sailing. Although a slight asymme- performances, although at different optimum 𝑐𝑟 ∕𝑑 val-
try is just perceivable for the data with AR = 5.14, the xflr5 ues. The accurate identification of this latter parame-
results of Fig. 14 indicate the superiority of straight foils ter is more critical for the rectangular geometries with
compared to swept ones (in terms of the lift curve slopes) for respect to the elliptic ones. Tapered geometries with
the relatively high aspect ratios typical of the appendages of moderate taper ratios are expected to perform inter-
dinghies and skiffs. The 𝐶𝐿 ∕𝐶𝐷 versus 𝛼 data (not reported) mediately between the rectangular and elliptic ones.
confirmed this superiority. Accordingly, so far the authors
have not found any solid indication about possible gains of • The rudder design needs to respond to two conflicting
requirements: low drag in navigation and delaying as

Castegnaro et al.: Preprint submitted to Elsevier Page 19 of 26


An Engingeering Method to Design Daggerboards and Rudders

much as possible the stall. It is sized as 15% of the • The race environmental conditions, in terms of:
lateral underwater surface (hull + appendages) and its expected True Wind Speed TWS, leg length 𝑙𝑅𝑒𝑔
upper part should present a chord reduction to cope (i.e., the distance between the upwind and down-
with the effects of the free water surface and of the wind buoys), expected wave period 𝑇1 , and sig-
hull boundary layer. /item A simple criterion, based nificant wave height 𝐻1∕3 ;
on evaluating the 𝐶𝑙 ∕𝐶𝑑 ratios at different operating • The hull characteristics: the length overall and
conditions, is proposed to preliminary select the 2D length at waterline (𝐿𝑂𝐴 and 𝐿𝑊 𝐿 ), the Bare
foil sections for the daggerboard and the rudder. Hull resistance curve 𝐷𝐵𝐻 = 𝑓1 (𝐵𝑆) as ob-
tained by software simulation (e.g., MaxSurf) or
Investigations have also been performed on the topic of the
by towing tank tests, the hydrostatic wet surface
sweep angle Λ, comparing xflr5 simulations with the results
𝑆𝐵𝐻 , and the volume displacement of the hull
obtained by equations from the literature to provide the lift
▽𝑐 ;
curve slope of swept rectangular and tapered foils. In partic-
ular: • the upwind and downwind sail dimensions (𝑆𝑆𝑈 𝑊
and 𝑆𝑆𝐷𝑊 , respectively), which are needed to
• Equation 24 provided by Kuchemann for swept rectan- compute the propulsive and side forces;
gular foils fairly agrees with the numerical results on • the force coefficients of the sails as function of
geometries with moderate sweep angles. On the con- the boat apparent wing angle awa’ (i.e., the an-
trary, Eq. 16 erroneously predicts the lift curve slope gle of the apparent wind with respect to the cen-
of tapered swept foils, both in terms of the maximum treline of the boat, see Fig. 4). The awa’ is the
𝐶𝐿 ∕𝛼 value and of the corresponding Λ. angle used in wind tunnel sail tests (where the
• For the relatively high aspect ratios typically used on model, being fixed, cannot experience the lee-
the foils of small racing yachts, the data indicates that way motion - see [35]), or the one which a wind
the highest 𝐶𝐿 ∕𝛼 is achieved with straight geometries vane (with the zero along the boat centreline)
(i.e., Λ = 0°), both for rectangular and tapered ones. would indicate during a real navigation. The ac-
tual apparent wind angle 𝐴𝑊 𝐴 (i.e., the angle of
the wind with respect to the boat motion, which
is identified by the direction of 𝐵𝑆 - see Fig. 4)
A. Benchmarking differs from awa’ of the leeway angle: 𝐴𝑊 𝐴 =
The underwater shapes of 20 commercial dinghies and awa’ + 𝛾. The sail force coefficients generally
skiffs are presented in Fig. 15, with the respective dimen- comes from wind tunnel tests or numerical sim-
sions (extracted from the lateral views) reported in table 8. ulations; according to the different authors (e.g.,
[3, 7, 28, 35]), either the lift and drag coefficients
𝐶𝐿𝑆 𝐶𝐷𝑆 or the driving and side force ones 𝐶𝑡𝑆
𝐶ℎ𝑆 are provided. Both coefficient combinations
B. VPP Algorithm and Equations
can be used, as long as the resulting forces are
Referring to Fig. 3, each block of the computation is projected along the direction of motion and the
entirely explained here in the following list (see the corre- perpendicular one (see the step nr. 6).
sponding numbers), which partly resembles the descriptive
• the boat apparent wind angle awa’ correspond-
one previously provided in Sec. 2. As a practical example,
ing to the sail force coefficients employed, for the
the results of the I, II, and last iteration are reported for the
upwind and downwind conditions, respectively
case involving a rectangular daggerboard with d = 1.3 m,
(or for other headings, in case of need).
𝑐𝑟 ∕𝑑 = 0.4 and the conditions reported in table 2.
• The polar characteristics of the 2D foil section
1. Several inputs are needed to start: at several Reynolds numbers, to account for the
performance variations with Re. Experimental
databases for symmetric airfoils can be obtained
Table 7 from the reports [15, 18, 19]; otherwise, for a
Computation of the daggerboard parameters in function of 𝑑 given foil section a database of polar character-
and of the 𝑘 ratio. Notice that the effective aspect ratio is istics at different Re could be built with the open-
twice the geometric one. source XFOIL software [20].
Planform Rectangular Tapered 1/4 Elliptic This input requires the selection of specific foil
𝑐𝑟 (1+𝜆) 1 sections both for the daggerboard and the rudder
𝑐𝑟 ⋅ 𝑑 ⋅𝑑 𝑐 ⋅𝑑
𝑆 [𝑚2 ] 2 4 𝑟
1 (see Sec. 5.1).
= 𝑘 ⋅ 𝑑2 = 0.5(1 + 𝜆)𝑘 ⋅ 𝑑 2 𝑘 ⋅ 𝑑2
4
𝐴𝑅𝐷𝐵 [-] 2. The design parameters for the daggerboard shall be
2/k 4
⋅ 1 8
given as input as well in step 2, and then modified
2
(= 2 𝑑𝑆 ) 1+𝜆 𝑘 𝜋⋅𝑘

in successive computations to assess different geome-


tries. These parameters are: the daggerboard draft (or

Castegnaro et al.: Preprint submitted to Elsevier Page 20 of 26


An Engingeering Method to Design Daggerboards and Rudders

Vaur
ien

Figure 15: Benchmarking on 20 commercial small crafts. Proportions not in scale.

semi-span) 𝑑 (in [m]) selected, the range of 𝑘 ratios planform (see table 7).
(= 𝑐𝑟 ∕𝑑) to assess during the computation, and the Each computation is performed on a single value of
generic planform shape (e.g., 1/4 of ellipse). All the 𝑘𝑖 and, once converged, it is then repeated with the
geometrical parameters of the daggerboard (surface 𝑘𝑖+1 one. When the computations on the entire 𝑘 range
𝑆𝐷𝐵 , aspect ratio 𝐴𝑅𝐷𝐵 , and chord length distribu- are completed, the results are recorded and the proce-
tion) are so defined in terms of 𝑘 and 𝑑 for a given

Castegnaro et al.: Preprint submitted to Elsevier Page 21 of 26


An Engingeering Method to Design Daggerboards and Rudders

dure applied again for different daggerboard drafts 𝑑 8. The Reynolds number is computed referring to the
and planforms for comparison. The 𝑘 range shall be mean chord length of the daggerboard 𝑐𝑚 :
selected so that the minimum 𝐴𝑅𝐷𝐵 remains higher
than 4, since the equations for finite wings with rela- 𝜌𝑊 ⋅ 𝐵𝑆 ⋅ 𝑐𝑚 𝜌 ⋅ 𝐵𝑆 ⋅ 𝑘 ⋅ 𝑑
𝑅𝑒𝐷𝐵 = ≈ 𝑊 (21)
tively high aspect ratios are not representative for lower 𝜇 𝜇
values [22, p.442].
where 𝜇 is the dynamic viscosity of the water.
3. Once the boat speed 𝐵𝑆 is imposed at the beginning
of the iteration, 𝐷𝐵𝐻 is computed from the resistance It can be seen from Eq. 21 that, for a given draft 𝑑,
curve of the bare hull (which can be approximated by the Reynolds number 𝑅𝑒𝐷𝐵 is approximately propor-
an interpolating polynomial, for instance); tional to the 𝑘𝑖 of that particular iteration. On the other
hand, at equal 𝑘, 𝑅𝑒 increases with 𝑑.
4. A leeway angle 𝛾 of first guess is assumed (typically,
𝛾1 = 0°); 9. For the daggerboard, the lift curve slope pertaining to
the selected foil section is identified according to the
5. The velocity triangles (see Fig. 4) are resolved with Reynolds number:
the 𝐵𝑆 and 𝛾 resulting from the previous iteration. In
particular, the following equations are used: 𝑎2𝐷 = 𝑓2 (𝑅𝑒𝐷𝐵 ) (22)

𝐴𝑊 𝐴 = 𝑎𝑤𝑎′ + 𝛾 (17) with 𝑓2 being an interpolation function fitting the data.


ℎ = 𝐵𝑆 ⋅ 𝑠𝑖𝑛(𝐴𝑊 𝐴) As shown in Fig. 5a for the 0012 and 64012 NACA
sections [18], the 𝑎2𝐷 parameter features important
𝛿 = 𝑎𝑟𝑐𝑠𝑖𝑛(ℎ∕𝑇 𝑊 𝑆) variations in the range 𝑅𝑒 < 3⋅106 , which is the range
𝜙 = 180 − (𝐴𝑊 𝐴 + 𝛿) of interest for most of the appendages [5, 8]. This
𝑇 𝑊 𝐴 = 180 − 𝜙 behaviour is due to thickening of the boundary layer
√ on the suction side of the foil section with decreas-
𝐴𝑊 𝑆 = 𝑇 𝑊 𝑆 2 + 𝐵𝑆 2 − 2𝐵𝑆 ⋅ 𝑇 𝑊 𝑆 ⋅ 𝑐𝑜𝑠(𝜙)
ing 𝑅𝑒 (a well known phenomenon in several techni-
cal disciplines, as turbomachinery design - see e.g.,
6. The propulsive and side forces 𝐹𝑃 and 𝐹𝑠 are com- [21, p.71]).
puted projecting the forces acting on the sails on the
direction of the motion (𝐵𝑆) and its perpendicular. If 10. The lift curve slope of the entire daggerboard 𝑎𝐷𝐵
the driving and heeling coefficients are used (see e.g., is computed according to the well-known equations
[7, p.175]), this gives: coming from the aerodynamics of wings (see [1, 22,
23]):
𝐹𝑡′ = 0.5𝜌𝑎𝑖𝑟 ⋅ 𝐴𝑊 𝑆 2 ⋅ 𝐶𝑡𝑆 ⋅ 𝑆𝑆 (18)
Elliptic planform:
𝐹ℎ′ = 0.5𝜌𝑎𝑖𝑟 ⋅ 𝐴𝑊 𝑆 2 ⋅ 𝐶ℎ𝑆 ⋅ 𝑆𝑆
𝜋 ⋅ 𝐴𝑅
𝐹𝑝 = 𝐹𝑡′ ⋅ 𝑐𝑜𝑠(𝛾) + 𝐹ℎ′ ⋅ 𝑠𝑖𝑛(𝛾) (19) 𝑎𝑒𝑙 = ( ) (23)
1 + 𝑎𝜋 𝐴𝑅
𝐹𝑠 = 𝐹ℎ′ ⋅ 𝑐𝑜𝑠(𝛾) − 𝐹𝑡′ ⋅ 𝑠𝑖𝑛(𝛾) 2𝐷

where 𝐹𝑡′ and 𝐹ℎ′ are the driving and heeling forces Rectangular planform:
along the boat centreline and the traversal direction, 𝑎2𝐷 ⋅ 𝑐𝑜𝑠Λ
respectively. According to the leg, the upwind and 𝑎𝑟𝑒 = √ (24)
( )
downwind sail surfaces 𝑆𝑆𝑈 𝑊 or 𝑆𝑆𝐷𝑊 are respec- 𝑎 𝑐𝑜𝑠Λ 2 𝑎2𝐷 𝑐𝑜𝑠Λ
1 + 2𝐷 +
tively used in Eq. 18. Whether the sail lift and drag 𝜋𝐴𝑅 𝜋𝐴𝑅
coefficients 𝐶𝐿𝑆 and 𝐶𝐷𝑆 are available instead (e.g.,
where Λ is the sweep angle [22, 23]. Since the sweep
[3, p.158]), they can be converted with the following
angle is generally included to introduce an elliptic load
equations:
distribution on a wing/foil which is not geometrically
𝐶ℎ𝑆 = 𝐶𝐿𝑆 ⋅ 𝑐𝑜𝑠(𝑎𝑤𝑎′ ) + 𝐶𝐷𝑆 ⋅ 𝑠𝑖𝑛(𝑎𝑤𝑎′ ) (20) elliptic [3, 10], this parameter is not needed in Eq. 23.

𝐶𝑡𝑆 = 𝐶𝐿𝑆 ⋅ 𝑠𝑖𝑛(𝑎𝑤𝑎′ ) − 𝐶𝐷𝑆 ⋅ 𝑐𝑜𝑠(𝑎𝑤𝑎′ ) Tapered wings with taper ratios 𝑐𝑟 ∕𝑐𝑡 ≈ 0.45 − 0.5
show lift curve slopes almost equal to the elliptic one
. [3], thus Eq. 23 can still be used with confidence.
For different taper ratios, Anderson [2] computes 𝑎𝑡𝑎𝑝
7. The Daggerboard geometrical parameters are computed multiplying the right-hand member of Eq. 23 by a fac-
(𝑆𝐷𝐵 , 𝐴𝑅𝐷𝐵 , and the mean chord 𝑐𝑚 ) from the inputs tor 𝑓 which is function of 𝑐𝑟 ∕𝑐𝑡 and the foil aspect ra-
provided at step nr. 2. As previously reported, the ef- tio. The plot of 𝑓 = (𝑐𝑟 ∕𝑐𝑡 , 𝐴𝑅) has not been reported
fective aspect ratio for the performance computation in this work, but the reader can refer to the original
of the daggerboard is doubled, according to the mir- freely-available NACA report [2], in case of need.
roring plane concept induced by the hull [3, 6, 10]

Castegnaro et al.: Preprint submitted to Elsevier Page 22 of 26


An Engingeering Method to Design Daggerboards and Rudders

11. The daggerboard lift coefficient 𝐶𝐿𝐷𝐵 required to bal- symmetric sections foresees what follows: the com-
ance the side force 𝐹𝑠 on the sails is computed as: putation of the minimum drag coefficient at zero lift
as
𝐹𝑠
𝐶𝐿𝐷𝐵 = (25) ( )
0.5𝜌𝑤 ⋅ 𝐵𝑆 2 ⋅ 𝑆𝐷𝐵 𝑅𝑒𝑟𝑒𝑓 0.11
𝐶𝑑𝑚𝑖𝑛 = (𝐶𝑑𝑚𝑖𝑛 )𝑟𝑒𝑓 ⋅ (33)
𝑅𝑒
12. The unviscid drag coefficient (related to the induced
velocity) 𝐶𝐷𝑖 is computed according to: where the subscript ref stands for the parameter at
𝑅𝑒𝑟𝑒𝑓 = 8 ⋅ 106 (such values can be found in the
𝐶𝐿2 report [15]). Alternatively, the decreasing trends of
𝐶𝐷𝑖 = (26) 𝐶𝑑𝑚𝑖𝑛 with 𝑅𝑒 presented in Fig. 5b for the NACA 0012
𝜋𝐴𝑅 and 64012 sections can be described by an interpolat-
Equation 26 is exact for elliptic planforms only, al- ing function 𝐶𝑑𝑚𝑖𝑛 = 𝑓3 (𝑅𝑒). The 𝐶𝑑 at different lift
though it can be applied to other planforms with a neg- coefficients (i.e., at different angles of attack) is then
ligible error [1, p.147]. computed as (see Fig. 16):
The daggerboard angle of attack (which is the leeway
𝐶𝑑 = 𝐶𝑑𝑚𝑖𝑛 + Δ𝐶𝑑
angle 𝛾 of the boat, when the iterations converge) is ( )
computed as: 𝐶𝑙
= 𝐶𝑑𝑚𝑖𝑛 + 𝑓4 (34)
𝐶𝑙𝑚𝑎𝑥
180 𝐶𝐿𝐷𝐵
𝛼= ⋅ (27)
𝜋 𝑎𝐷𝐵
with the function 𝑓4 represented in Fig. 16, 𝐶𝑙 =
𝐶𝐿𝐷𝐵 , and both 𝐶𝑑𝑚𝑖𝑛 and 𝐶𝑙𝑚𝑎𝑥 taken at the corre-
In addition, the local angle of attack 𝛼2𝐷 might be
sponding 𝑅𝑒 (see Fig. 5 and table 1). The maximum
needed to compute the viscous drag from the foil sec-
lift coefficient of different airfoils at several Reynolds
tion polar at the specific 𝑅𝑒; 𝛼2𝐷 is computed by the
can be found in [15], or in [18] for the 0012 and 64012
following equations (see [1]).
sections (Fig. 5c).
Elliptic Planform:
( 𝐶𝐿𝐷𝐵 )
180
𝛼2𝐷 = 𝛼 − ⋅ (28)
𝜋 𝜋 ⋅ 𝐴𝑅𝐷𝐵

Rectangular Planforms (see [1, p.147]):

𝐶𝐿𝐷𝐵
𝐴1𝑟𝑒 = (29)
𝜋 ⋅ 𝐴𝑅𝐷𝐵
𝜇𝑟𝑒 = 2𝜋 ⋅ 𝑘𝑖 ∕8 (30)
( )
𝜇𝑟𝑒 ⋅ 𝛼[𝑟𝑎𝑑] 𝜋
𝜏 = (1∕𝜇𝑟𝑒 ) ⋅ − − 1 (31)
𝐴1𝑟𝑒 4
( 𝐶𝐿𝐷𝐵 )
180
𝛼2𝐷 = 𝛼 − ⋅ ⋅ (1 + 𝜏) (32)
𝜋 𝜋 ⋅ 𝐴𝑅𝐷𝐵 Figure 16: The concept of generalized polar allows to
compute the increase of the drag coefficient Δ𝐶𝑑 with
The angle 𝛼2𝐷 is constant for wings with elliptical respect to the zero lift condition at any Reynolds num-
loading distribution, whereas it reduces towards the ber; graph adapted from [15].
tip for the rectangular geometries because of the local
increase of downwash (see [1, 10]). Accordingly, for
rectangular foils the angles computed with Eq.s 27 - 14. The overall daggerboard drag is computed as sum of
32 should be integrated and averaged along the entire the viscous and non-viscous contributions:
foil semispan, although a direct use of Eq.s 27 - 32
results in an acceptable approximation [1]. 𝐶𝐷𝐷𝐵 = 𝐶𝑑 + 𝐶𝐷𝑖 (35)
13. If the foil section polars at different 𝑅𝑒 are not directly 1
𝐷𝐷𝐵 = 𝜌 𝐶 ⋅ 𝐵𝑆 2 ⋅ 𝑆𝐷𝐵 (36)
available, the viscous drag coefficient 𝐶𝑑 can be com- 2 𝑤 𝐷𝐷𝐵
puted according to the generalized section polar pro-
posed by Jacobs and Sherman [15], which allows an 15. As anticipated, the rudder is sized according to a rule
estimation of the polar of the generic section at differ- of thumb suggested by Trower for dinghies [5, p.113],
ent Reynolds numbers. The application to the case of rather than according to fluid-dynamics equations. In

Castegnaro et al.: Preprint submitted to Elsevier Page 23 of 26


An Engingeering Method to Design Daggerboards and Rudders

particular, the rudder surface 𝑆𝑅 is assumed to be equal 18. At the given 𝐵𝑆, the entire drag of the hull with the
to the 15% of the underwater lateral plane (i.e., the appendages (𝐷𝑇 𝑂𝑇 = 𝑓 (𝐵𝑆)) is then computed as
lateral projection of the underwater surface 𝑆𝐵𝐻𝑙𝑎𝑡 +
𝑆𝐷𝐵 + 𝑆𝑅 ), giving: 𝐷𝑇 𝑂𝑇 = 𝐷𝐵𝐻 + 𝐷𝐷𝐵 + 𝐷𝑅 (+𝐷𝐴𝑊 ) (40)

𝑆𝑅 ≈ 0.176 ⋅ (𝑆𝐵𝐻𝑙𝑎𝑡 + 𝑆𝐷𝐵 ) (37) with the addition of 𝐷𝐴𝑊 only for the upwind leg.
19. Finally, the initially guessed boat speed value 𝐵𝑆 is
The positioning of the daggerboard with respect to the verified comparing the propulsive force on the sails
rudder and the centre of force on the sails [5, 7]) does 𝐹𝑝 with the total drag 𝐷𝑇 𝑂𝑇 . In particular, 𝐵𝑆 is in-
not affect the performance estimation. Nonetheless, creased if
the boat centering has to be verified (see Sec. 3) to
achieve the sailing balance before finalizing the foil 𝐷𝑇 𝑂𝑇 < 𝐹𝑃
design.
or decreased if
16. Since the rudder is sized for maneuvering and not to
contribute in resisting to leeway, at this step the sail- 𝐷𝑇 𝑂𝑇 > 𝐹𝑃
ing balance (i.e., the equilibrium between turning mo-
ments along the yaw axis of the boat) is assumed to be until converging to a 𝐵𝑆 value such that 𝐷𝑇 𝑂𝑇 =
achieved. This involves that the drag of the rudder dur- 𝐹𝑃 . At each new iteration the leeway angle previously
ing navigation (unless if used to correct the trajectory) computed returns to step 4 and it updates the compu-
is only the viscous one around the zero-lift operation: tation of the velocity triangles.
20. When the iterations are converged, the daggerboard
1 angle of attack 𝛼 (computed with Eq. 27) actually is
𝐷𝑅 = 𝜌𝑤 𝐵𝑆 2 ⋅ 𝐶𝑑𝑚𝑖𝑛 ⋅ 𝑆𝑅 (38)
2 the leeway angle of the boat 𝛾.
where 𝐶𝑑𝑚𝑖𝑛 is function of the Reynolds number and 21. The Velocity Made Good VMG is computed as:
of the foil section selected for the rudder. According
to the data reported in [26], the decrease of rudder 𝑉 𝑀𝐺 = 𝐵𝑆 ⋅ |𝑐𝑜𝑠(𝑇 𝑊 𝐴)| (41)
drag caused by the daggerboard wake and downwash
is Δ𝐷𝑅 < 11%. This makes the computation of 𝐷𝑅 22. The times 𝑡𝑈 𝑊 and 𝑡𝐷𝑊 are finally calculated for the
with Eq. 38 an acceptable approximation even for rud- upwind and downwind legs respectively, dividing the
ders positioned in line behind the daggerboard. distance between bouys 𝑙𝑅𝑒𝑔 for the VMG:

17. During the upwind navigation an additional drag is 𝑙𝑅𝑒𝑔


caused by the boat interaction with the waves natu- 𝑡= (42)
𝑉 𝑀𝐺
rally existing on the race field. Sailing windward, the
boat heads against the approaching waves, with the and the overall time is given as sum of the two:
bow heading up and down in a pitching motion with a
slowing effect; this is called added resistance in waves 𝑡𝑡𝑜𝑡 = 𝑡𝑈 𝑊 + 𝑡𝐷𝑊 (43)
and its modeling is rather complex. In this work the This latter parameter is then used to identify the best
equation proposed by the researchers of Delft [27] and appendage combination: the lower the time 𝑡𝑡𝑜𝑡 , the
tabulated in [3] is used: quicker will be the boat during the race.
2 [ ]𝑏 It shall be remarked that the 𝑡 values so computed do
𝜌𝑤 𝑔 ⋅ 𝐿𝑊 𝐿 ⋅ 𝐻1∕3 ▽
1∕3
𝑘𝑦𝑦
𝐷𝐴𝑊 = 𝑎 ⋅ 10 ⋅ 𝑐
2 not represent actual racing times, since they cannot ac-
102 𝐿𝑊 𝐿 𝐿𝑊 𝐿 count for: i) the time needed for tacking and gybing,
(39) and ii) the slowing action induced by the rudder acti-
vation (for instance to correct the route during upwind
where 𝑎 and 𝑏 are two coefficients tabulated in func- navigation or to bring back the boat to a flat attitude
tion of wave mean period 𝑇1 and the Froude number in downwind conditions).
𝐹𝑛 for the different heading angles (see [3, p.93].), 𝑔 is
the gravitational acceleration, and 𝑘𝑦𝑦 is the gyradius
(a parameter related to the moment of inertia of the CRediT authorship contribution statement
boat along the transversal axis 𝑦). According to Lars-
S. Castegnaro: Methodology, Conceptualization, Inves-
son and Eliasson [3, p.93], taking 𝑘𝑦𝑦 = 0.25 ⋅ 𝐿𝑂𝐴 is
tigation, Writing - Original draft preparation. C. Battisti:
a reasonable assumption and this appears appropriate
Conceptualization, Methodology, Validation. M. Poli: Data
within the scope of this work.
Curation. A. Lazzaretto: Writing - Reviewing and Editing,
Project administration.

Castegnaro et al.: Preprint submitted to Elsevier Page 24 of 26


An Engingeering Method to Design Daggerboards and Rudders

1200 References
1100 [1] Hermann Glauert. The elements of aerofoil and airscrew theory.
Cambridge University Press, 1983.
1000 [2] Raymond F Anderson. Determination of the characteristics of tapered
tuw [s]

900
wings. Technical report, NATIONAL ADVISORY COMMITTEE
FOR AERONAUTICS, LANGLEY FIELD, VA, 1936.
800 [3] Rolf Eliasson and Larsson. Principles of yacht design. McGraw Hill,
ellip. (4 m/s)
ellip. (6 m/s) 2007.
700
rect. (4 m/s) [4] Howard Irving Chapelle. American small sailing craft, their design,
rect. (6 m/s)
600 development, and construction. WW Norton & Company, 1951.
0 0.2 0.4 0.6 0.8 1
cr/d [5] Gordon Trower. Yacht and Small Craft Design: From Principles to
Practice. Helmsman Books, 1992.
(a) [6] Sergio Crepaz. Teoria e progetto di imbarcazioni a vela. Zanichelli,
1986.
1100 [7] Fabio Vittorio Fossati. Teoria dello yacht a vela. Polipress, 2007.
[8] David Vacanti. Keel and rudder design. Professional BoatBuilder,
1000
95:6, 2005.
[9] Ross Garrett. The symmetry of sailing: the physics of sailing for
900
yachtsmen. Sheridan House, Inc., 1996.
tdw [s]

[10] C Marchaj. Aero-Hydrodynamics of Sailing. 1979. U. Mursia Editore


800
(italian version), 1987.
ellip. (4 m/s)
[11] T. Acerbi, R. Andersson, E. Eriksson, S. Granli, E. Jacobs, F. Rita,
700 ellip. (6 m/s)
rect. (4 m/s) R. Sahlberg, and E. Werner. Chalmers formula sailing. high perfor-
rect. (6 m/s) mance skiff, marine design project. Chalmers University of Tech-
600
0 0.2 0.4 0.6 0.8 1 nology, Chalmers Reproservice, 2017. https://offshorevast.se/
cr/d
media/2017/12/Formula_Sailing_Reduced-002.pdf (last accessed: June

(b) 1 2020).
[12] U Cella, Me Biancolini, C Alberto, and F Francesco. Tecnologie cae
2200 avanzate per la progettazione delle “foil” di un catamarano classe a.
Analisi & Calcolo, 2018.
2000 [13] A Cirello and A Mancuso. A numerical approach to the keel design
of a sailing yacht. Ocean engineering, 35(14-15):1439–1447, 2008.
1800 [14] Anthony F Molland, Stephen R Turnock, and Dominic A Hudson.
ttot [s]

Ship resistance and propulsion. Cambridge university press, 2017.


1600 [15] Eastman N Jacobs and Albert Sherman. Airfoil section characteris-
ellip. (4 m/s) tics as affected by variations of the reynolds number. NACA report,
ellip. (6 m/s)
1400
rect. (4 m/s)
586:227–264, 1937.
rect. (6 m/s) [16] A Deperrois. Xflr5 analysis of foils and wings operating at low
1200
0 0.2 0.4 0.6 0.8 1 reynolds numbers. 2009. https://engineering.purdue.edu/~aerodyn/
cr/d AAE333/FALL10/HOMEWORKS/HW13/XFLR5_v6.01_Beta_Win32%282%29/
Release/Guidelines.pdf (last accessed: June 1 2020).
(c) [17] Philippe Berman. Keel vs dagger-boards. which is better for a sailing
cat? Boating New Zealand, 10:., 2018.
Figure 17: 𝑡𝑈 𝑊 , 𝑡𝐷𝑊 , and 𝑡𝑇 𝑂𝑇 trends for different environ- [18] Laurence K Loftin Jr and Hamilton A Smith. Aerodynamic character-
istics of 15 naca airfoil sections at seven reynolds numbers from 0.7
mental conditions with respect to those previously analysed:
′ x 10 (exp 6) to 9.0 x 10 (exp 6). Technical report, NACA, 1949.
TWS = 4 m/s, 𝐻1∕3 = 0.3 m and 𝑇1 = 4 s; TWS = 6 m/s, [19] James C Ferris, Robert J McGhee, and Richard W Barnwell. Low-

𝐻1∕3 = 0.3 m and 𝑇1 = 2.5 s. speed wind-tunnel results for symmetrical nasa ls (1)-0013 airfoil.
1987.
[20] Mark Drela. Xfoil: An analysis and design system for low reynolds
number airfoils. In Low Reynolds number aerodynamics, pages 1–12.
Acknowledgement Springer, 1989.
[21] R Allan Wallis. Axial flow fans. Design and Practice. Academic
We dedicate this work to Carlo Gomiero: engineer, friend, Press, 1961.
and mate, who structurally designed and built several of our [22] John D Anderson. Fundamentals of Aerodynamics. McGraw Hill,
appendages. We thank Federica Feruglio for her meticu- 2007.
lous research of data on commercial boats and dr. Fabrizio [23] D Kuchemann. The aerodynamic design of aircraft. Pergamon, Lon-
don, 1978.
Medeossi for the surface roughness measurements. We fi- [24] Thomas H Carolus and Ralf Starzmann. An aerodynamic design
nally acknowledge all the students, professors, and volun- methodology for low pressure axial fans with integrated airfoil polar
teers who have provided, and currently provide, support to prediction. In ASME 2011 Turbo Expo: Turbine Technical Conference
the Metis Team of the University of Padova, Italy. The ac- and Exposition, pages 335–342. American Society of Mechanical En-
tivities of the Metis Team are supported by the University of gineers Digital Collection, 2011.
[25] JA Keuning, Michiel Katgert, and Kees Jan Vermeulen. Keel-rudder
Padova since 2008. interaction on a sailing yacht. In 19th International HISWA Sym-
posium on Yacht Design and Yacht Construction, Amsterdam, The
Netherlands, 2006.
[26] Daniele Torriani. Progettazione della deriva e del timone di arete’ per

Castegnaro et al.: Preprint submitted to Elsevier Page 25 of 26


An Engingeering Method to Design Daggerboards and Rudders

le regate “1001 vela cup” (in italian). (design of the daggerboard and ECOS2017 Conference, San Diego, USA. Author or co-author of more than
rudder of aretè for the 1001 vela cup regatta), 2012. 230 papers published in international journals or conference proceedings.
[27] JA Keuning and Uco B Sonnenberg. Approximation of the hydro- He is responsible of Metis, the student sailing team of the University of
dynamic forces on a sailing yacht based on the’delft systematic yacht Padova, since 2008; he shares with the other authors a deep passion for
hull series’. Delft University of Technology, Faculty of Mechanical, sailing since youth, and participated also to national and international sail-
Maritime, and Materials Engineering, 1998. ing races,
[28] Andrea Carradori. Analisi numerica dell’aerodinamica di differenti
piani velici per un’imbarcazione da regata (in italian). (a numerical
analysis of the aerodynamics of different sail-plans for a racing sail-
boat), 2016.
[29] Linda Wolstenholme. Sailing close to the statistics. Significance,
6(2):52–57, 2009.
[30] Mark Drela and Michael B Giles. Viscous-inviscid analysis of tran-
sonic and low reynolds number airfoils. AIAA journal, 25(10):1347–
1355, 1987.
[31] TE Tinling and Richard W Kolk. The effects of mach number and
reynolds number on the aerodynamic characteristics of several 12-
percent-thick wings having 35 of sweepback and various amounts of
camber. NACA RM-A50K27, 1951.
[32] Louis S Stivers, Ira H Abbott, and Albert E von Doenhoff. Summary
of airfoil data, report no. 824. Technical report, NACA, 1945.
[33] Ted Brewer. Fin keels are here to stay. Cruising world, .:100–103,
1981.
[34] Dieter Scholz, Ole BÖTTGER, and Bernd TRAHMER. Short Course
Aircraft Design. DGLR, 2007.
[35] Ignazio Maria Viola. Downwind sail aerodynamics: a cfd investiga-
tion with high grid resolution. Ocean engineering, 36(12-13):974–
984, 2009.
Stefano Castegnaro is project engineer at Area Impianti, Italy. Formerly
post-doc at the von Karman Institute, he received two master degrees: one
in Industrial Design at the University IUAV of Venice, Italy and one in
Aerospace Engineering at the University of Padova, Italy where he also ob-
tained the PhD in Industrial Engineering in 2018. He was visiting student at
the Universidad Politecnica de Madrid, Spain within the EU Erasmus Pro-
gram. Before the PhD he worked in the nautical production field (REA Ma-
rine and Slyder Yachts), participating to the design and construction of some
innovative sailing yachts. He joined the VKI after a one-year-experience at
the italian CERN Industrial Liason Office of the National Institute for Nu-
clear Physics. His research activity regards turbomachinery aerodynamics
and related testing rigs; he is a passionate sailor.
Cristiano Battisti graduated in 1996 at Rome University in Aeronautical
Engineering. As a specialist in fluid dynamics and structural design for
metallic and advanced composite materials, he worked since graduation in
the marine sector. He combined a life-long sailing passion with profes-
sional experience with the most successful shipyards. His work experience
ranges from small recreational craft and workboats to large carbon com-
posite sailing megayacht, and up to cruise and ferry vessels. Since 2008
he joined Padova University design team for successful development and
construction of bio-composite racing dinghies. He is currently working for
Fincantieri shipyard as Head of the structural design department for large
cruise vessels.
Matteo Poli graduated at the University of Padova in 2013 in aerospace en-
gineering. His heterogeneous interests and fields of expertise range from
sailing boats to rocket propulsion. For five years he participated to the de-
sign and construction of racing skiffs at the University of Padova, with wood
and bio-composites. He later created a student group for testing and launch-
ing sounding rockets. Passionate sailor, he currently works at Airbus De-
fence and Space as Assembly, Integration and Test engineer in Ottobrunn,
Germany.
Andrea Lazzaretto is professor of Energy Systems at the University of Padova,
where he was first associate professor in Design of Turbomachinery and as-
sistant professor in Fluid Machines. PhD in Energy Engineering in 1992
and Research Scholar at the Tennessee Tech University, USA, in 1991.
ASME fellow since 2010, member of the Executive Committee of the ASME
Advanced Energy System Division since 2011. Winner of prestigious awards,
among which the ASME E. F. Obert Award for outstanding papers on Ther-
modynamics in 1998, 2007 and 2018 and the Best paper Award of the

Castegnaro et al.: Preprint submitted to Elsevier Page 26 of 26


Table 8: Benchmarking Analysis on commercial dinghies and skiffs.
Mast
𝐿𝑂𝐴 [m] 𝐿𝑊 𝐿 [m] Beam [m] 𝑆𝑆𝑈 𝑊 [𝑚2 ] d [m] 𝑑𝑅 [𝑚] 𝑐𝑟 [m] 𝑐𝑟𝑅 [m] (𝑐𝑟 ∕𝑑)𝐷𝐵 [-] (𝑐𝑟 ∕𝑑)𝑅 [-] 𝑆𝐷𝐵 [𝑚2 ] 𝑆𝑅 [𝑚2 ] 𝑆𝐷𝐵 ∕𝑆𝑅 𝑆𝑡𝑜𝑡 [𝑚2 ]
Height [m]
Flying Junior 4.04 - 1.5 - 9.7 0.86 0.63 0.51 0.24 0.59 0.381 0.29 0.15 52% 0.44
420 4.2 - 1.63 6.26 10.25 0.82 0.66 0.45 0.26 0.55 0.394 0.37 0.17 46% 0.54
Vaurien 4.08 - 1.47 - 10.8 0.88 0.56 0.31 0.23 0.35 0.411 0.27 0.13 48% 0.4

Castegnaro et al.: Preprint submitted to Elsevier


Snipe 4.72 4.11 1.52 - 10.8 0.85 0.6 0.58 0.28 0.68 0.467 0.37 0.17 46% 0.54
Skipper 4 - 1.6 - 11 0.74 0.37 0.31 0.19 0.42 0.514 0.23 0.07 30% 0.3
29er 4.45 4.24 1.77 6.25 12.5 0.73 0.62 0.28 0.2 0.38 0.323 0.20 0.12 60% 0.32
470 4.7 4.4 1.69 - 12.7 0.93 0.55 0.45 0.23 0.48 0.418 0.33 0.13 39% 0.46
Tridente 14 4.3 4.3 1.7 - 13 0.86 0.77 0.34 0.26 0.40 0.338 0.23 0.16 70% 0.39
Fireball 4.93 - 1.37 6.8 13.3 1.1 0.7 0.4 0.23 0.36 0.329 0.33 0.16 48% 0.49
Strale 4.9 4.35 1.57 - 13.5 0.96 0.56 0.45 0.26 0.47 0.464 0.34 0.11 32% 0.45
29erXX 4.45 4.24 1.77 6.25 15 0.79 0.53 0.32 0.22 0.41 0.415 0.25 0.12 48% 0.37
Yngling 6.37 4.7 1.7 - 15.83 0.76 0.59 0.94 0.39 1.24 0.661 0.42 0.19 45% 0.61
Lightning 5.8 - 1.98 7.98 16.4 1.2 0.57 0.57 0.3 0.48 0.526 0.51 0.17 33% 0.68
505 5.05 - 1.88 - 17.5 1.45 0.59 0.42 0.19 0.29 0.322 0.48 0.11 23% 0.59
Flying Scot 5.8 5.6 2 8.6 17.65 1.04 0.69 0.46 0.35 0.44 0.507 0.48 0.24 50% 0.72
Flying Dutchman 6.06 5.5 1.78 - 18.6 1.03 0.83 0.5 0.36 0.49 0.434 0.43 0.25 58% 0.68
49erFX 4.99 - 2.9 7.5 19.6 1 0.75 0.42 0.31 0.42 0.413 0.35 0.23 66% 0.58
49er 4.995 - 1.69 8.1 21.2 0.97 0.77 0.4 0.28 0.41 0.364 0.36 0.2 56% 0.56
Tempest 6.66 - 1.92 - 23.11 0.82 0.76 0.66 0.35 0.80 0.461 0.54 0.29 54% 0.83
An Engingeering Method to Design Daggerboards and Rudders

Elliott 6 m 6 6 2.35 - 23.6 1.3 0.85 0.28 0.28 0.22 0.329 0.36 0.22 61% 0.58
Star 6.922 4.724 1.734 9.652 26.5 0.79 0.79 1.41 0.41 1.78 0.519 0.93 0.43 46% 1.36

Page 27 of 26
View publication stats
Castegnaro et al.: Preprint submitted to Elsevier
Table 9: For the case presented in Sec. 3, the results of the first (I), second (II) and last (n) iteration are reported for the rectangular foil with d = 1.3 m.
𝑅𝑒𝐷𝐵
iter. 𝑐𝑟 ∕𝑑 d [m] 𝐵𝑆𝑖 [m/s] 𝛾𝑖 [°] 𝐷𝐵𝐻 [N] AWS [m/s] TWA [°] 𝐹𝑃 [N] 𝐹𝑠 [N] 𝑎2𝐷 [1/rad] 𝑎𝐷𝐵 [1/rad] 𝐶𝐿𝐷𝐵 𝐶𝐷𝑖 𝐶𝑑 𝐶𝑑𝑅 𝐷𝐷𝐵 [N] 𝐷𝑅 [N] 𝐷𝐴𝑊 [N]
106
I 0.4 1.3 4.0 0.0 255.2 8.3 44.8 265.4 789.3 2.4 3.95 5.59 0.143 0.001 0.007 0.006 44 9.6 43.4
II 0.4 1.3 3.5 2.0 216.0 7.9 45.5 260.5 693.8 2.1 3.90 5.49 0.164 0.002 0.007 0.006 36 7.4 41.9
n 0.4 1.3 3.1 2.9 170.6 7.5 44.6 247.3 628.4 1.8 3.86 5.41 0.193 0.002 0.007 0.006 30 5.7 40.2
iter. 𝐷𝑇 𝑂𝑇 [N] 𝐹𝑃 − 𝐷𝑇 𝑂𝑇 [N] VMG [m/s] t [s]
I 352.4 -87.0 2.8 704
II 301.2 -40.6 2.5 816
n 246.5 0.8 2.2 915
An Engingeering Method to Design Daggerboards and Rudders

Page 28 of 26

You might also like