scieee Science in your language
[en] (orig)

Simulation-driven design of sailing yachts and motor boats

Abstract

The design of yachts and boats can significantly benefit from simulation-driven design (SDD) using codes of Computational Fluid Dynamics (CFD). In SDD a large number of virtual prototypes is in- vestigated numerically for key objectives. In hydro- and aerodynamics objectives often relate to resistance and lift which govern the performance of both sailing yachts and fast motor boats. In order to reduce the dimensionality of the design space, i.e., the degrees-of-freedom, a parametric approach is utilized. For the flow simulation different levels of fidelity are used, ranging from potential flow analysis to viscous flow simulation solving the RANS equations. Design examples applying the SDD approach will be presented for both a sailing yacht and a motor boat. The sailing yacht is a 20m catamaran for worldwide travel and the motor boat is a 6m planing boat for day cruises. Parametric models for the two vessels will be discussed, comprising the generation of surfaces and watertight tri-meshes, the latter of which can be fed to the CFD code of choice. Here SHIPFLOW® and FINE™/Marine were applied in connection with CAESES® which provided both the shapes and the integration of CFD for SDD. To close the simulation driven design cycle of the sailing catamaran an appended version of the parametric model with rudders and daggerboard is used for virtual tank testing. Combining these results with a suitable sail model allows for an accurate velocity prediction (VPP) in an early design stage.

Read accessible full text

Simulation-driven design of sailing yachts and motor boats

Author: Hasubek, Bodo,Harries, Stefan
Publisher: CIMNE
Year: 2017
Source: https://upcommons.upc.edu/bitstream/2117/331057/1/Marine-2017-25_Simulation-driven%20design.pdf
Simula ion-d i en design o sailing yach s and mo o boa s
VII In e na ional Con e ence on Compu a ional Me hods in Ma ine Enginee ing
MARINE 2017
M. Visonneau, P. Queu ey and D. Le Touzé (Eds)
SIMULATION-DRIVEN DESIGN OF SAILING YACHTS AND MOTOR
BOATS
Bodo Hasubek∗, S e an Ha ies†
∗D eamca che One
Sonnenblumen ing 39, 86415 Me ing, Ge many
[email p o ec ed], www.d eamca che one.de
†FRIENDSHIP SYSTEMS AG
Benzs . 2, 14482 Po sam, Ge many
[email p o ec ed], www. iendship-sys ems.com
Key wo ds: Simula ion-D i en Design (SDD), Yach Design, Compu a ional Fluid Dynamics (CFD), Pa a-
me ic Modeling, Op imiza ion
Abs ac . The design o yach s and boa s can signi ican ly bene i om simula ion-d i en design (SDD)
using codes o Compu a ional Fluid Dynamics (CFD). In SDD a la ge numbe o i ual p o o ypes is in-
es iga ed nume ically o key objec i es. In hyd o- and ae odynamics objec i es o en ela e o esis ance
and li which go e n he pe o mance o bo h sailing yach s and as mo o boa s. In o de o educe he
dimensionali y o he design space, i.e., he deg ees-o - eedom, a pa ame ic app oach is u ilized. Fo he
low simula ion di e en le els o ideli y a e used, anging om po en ial low analysis o iscous low
simula ion sol ing he RANS equa ions. Design examples applying he SDD app oach will be p esen ed o
bo h a sailing yach and a mo o boa . The sailing yach is a 20m ca ama an o wo ldwide a el and he
mo o boa is a 6m planing boa o day c uises. Pa ame ic models o he wo essels will be discussed,
comp ising he gene a ion o su aces and wa e igh i-meshes, he la e o which can be ed o he CFD
code o choice. He e SHIPFLOW®and FINE™/Ma ine we e applied in connec ion wi h CAESES®which
p o ided bo h he shapes and he in eg a ion o CFD o SDD. To close he simula ion d i en design cycle
o he sailing ca ama an an appended e sion o he pa ame ic model wi h udde s and dagge boa d is used
o i ual ank es ing. Combining hese esul s wi h a sui able sail model allows o an accu a e eloci y
p edic ion (VPP) in an ea ly design s age.
1 In oduc ion
While CFD ship hull imp o emen s and op imiza ions a e qui e common o la ge comme cial essels,
CFD d i en design o saing yach s and small mo o boa s is usually limi ed o mul i-million dolla p ojec s
such as he Ame ica’s Cup. Cos s a e assumed o be p ohibi i e and incen i es a e missing. Fuel consump ion
is no an issue o pleasu e boa s and sailing pe o mance is a ibu ed o a “hea y” design o o he lack o
abili ies o he skippe . Howe e , ecen de elopmen s in pa ame ic design o ship hulls and a o dable CFD
compu a ions ha e changed he playing ield. In his pape he in es iga ion o a new design o a 20m sailing
mul ihull (Fig. 1) is p esen ed and he classic Ri a Junio as planing c a is e isi ed. Bo h designs ha e
1
324
Bodo Hasubek and S e an Ha ies
in common ha hey use a ha d chined hull. Ne e heless, he me hods discussed can be eadily applied o
ound-bilge hulls as well.
Figu e 1: D eamCa che One
Fo he sailing ca ama an ha d chines we e chosen o easy
manu ac u ing in aluminium based on de elopable su aces. I
is shown ha a sys ema ic a ia ion o he hull pa ame e s leads
o a signi ican educ ion o o al esis ance compa ed o a man-
ually lo ed hull. A e de ining he op imiza ion goal, a ully
pa ame ic model is de eloped o he canoe body o a single
hull using CAESES®. Fo he design selec ion a mul iple s age
app oach was aken: 4508 models we e in es iga ed using he
po en ial low code o SHIPFLOW® o wa e esis ance. Sub-
sequen ly, he mos p omising pa ame e anges we e chosen
o analyse 521 models wi h ega d o hei o al esis ance us-
ing SHIPFLOW®’s zonal app oach which combines a po en-
ial low solu ion o he wa e esis ance wi h a iscous esis-
ance solu ion o he ea hal o he hull an he wake. The bes
design was selec ed and e i ied by compa ing he esis ance
cu es o he whole ope a ing ange using bo h FLOWTECH’s
SHIPFLOW®and NUMECA’s FINE™/Ma ine. The pa ame -
ic model hen is e ined o include wo hulls as well as ud-
de s and dagge boa ds. This model is used o ind a sui able
dis ance be ween he hulls and o gene a e i ual ank es ing
inpu s o a eloci y p edic ion p og am (VPP) unde di e en
sailing condi ions using FINE™/Ma ine.
The example o he mo o boa a planing speed p esen s
an au oma ic minimiza ion o esis ance ia a design-o -
expe imen (DoE) combined wi h a de e minis ic sea ch. CAESES®and FINE™/Ma ine we e coupled as
an in eg a ed solu ion (Fig. 2). The example se es o highligh he p ocedu e and he a ious elemen s in-
ol ed – namely a iable geome y, high- ideli y CFD and o mal op imiza ion – along wi h imp o emen s
ha could be achie ed.
2 Simula ion-D i en Design o a sailing ca ama an
2.1 Design Objec i es
Figu e 2: Fas mo o boa modeled wi h
CAESES®
Fo mo o d i en essels he design objec i es a e usually
p o ided by he clien in e ms o pay load and eloci y e-
qui emen s. A sailing yach , howe e , ope a es a an a bi a iy
numbe o load and eloci y condi ions which canno be eas-
ily p edic ed. The e o e, o he load condi ion he wo s case
scena io o “ ully loaded” was chosen a a displacemen o 36 .
Due o wa e heo y he o al esis ance s. F oude numbe
cu e shows a ce ain “hump” a a ound Fn=0.3. This hump
can be seen in he esis ance cu e o he e e ence design. Fo
he ca ama an wi h a wa e line leng h o 20 me e s Fn=0.3
ansla es in o a boa speed o 8.2 kno s. Appa en ly, he speed
2
325
Bodo Hasubek and S e an Ha ies
ange be ween 8 and 10 kno s should be wi hin each o a ca ama an o his size e en in ligh winds. The e-
o e, he educ ion o his hump was chosen as he p ima y op imiza ion goal. A he same ime he op imiza-
ion should no sac i ice he speed abili y in no mal wind condi ions. F om analy ical es ima es conside ing
an upwind sail a ea o 270m2 he no mal upwind abili y esul s in a ound 12 o 13 kno s. This leads o a wo
speeds op imiza ion app oach o Fn=0.3 (8.2 kno s) and Fn=0.44 (12 kno s). Consequen ly, all 4508
models we e e alua ed o hese wo speeds o selec he designs wi h he bes o e all pe o mance.
2.2 Pa ame ic Models
The pa ame ic models use h ee sec ions, an a , a main and a bow sec ion. A model wi h h ee pla es
and one wi h ou pla es below he wa e line was de eloped o in es iga e whe he a be e app oxima ion
o an op imum ellip ical shape by using mo e pla es would lead o be e designs. Bo h sec ion models use
gi en, bu a iable wa e line and keel line cu es as well as a la e angle and a dead ise angle cu e o e
he comple e hull leng h. This allows o a lexible posi ioning o he main sec ions. The h ee pla e model
u he uses he sec ional a eas as inpu o gene a e he sec ion.
Figu e 3: Th ee pla e model: sec ion geome y Figu e 4: Fou pla e model: sec ion geome y
Fo educ ion o o al esis ance he we ed su ace a ea needs o be conside ed as well. In he cu en
app oach he leng h o he sec ion cu e (leng h om p1-p2-p3-p4in Fig. 3 ) was minimized using a quad a ic
equa ion o ind he posi ions o poin s p2and p3in dependance o he sec ional a ea, dead ise angle, he
side angle and he la e angle. Using a B en minimiza ion app oach wi hin CAESES® he side angle was
a ied o minimize he leng h p1-p2-p3-p4. The idea ollows he assump ion ha i he cu e leng hs o he
sec ions a e minimal and he ha d chines connec ing he poin s among he a , main and bow sec ions a e
ai , hen he we ed a ea o he su aces c ea ed be ween hese ha d chines is a leas close o i s minimum
as well. Analysing he c ea ed models showed ha he we ed su ace a ea among all designs a ied only by
a maximum o 1.5%.
Many hand-made design a ia ions su e om a ying displacemen alues [1]. In he cu en pa ame ic
app oach he main sec ion a ea is a ied using a second op imiza ion o mee he displacemen o 18 pe hull
wi h a de ia ion o less han 0.1‰. By keeping he displacemen cons an any a ia ions in esis ance can be
di ec ly a ibu ed o he changes in he geome y.
The ou pla e model (Fig. 4) ollows a simila app oach whe eby he geome y depends on 1=Tc, 2,
3
326
Bodo Hasubek and S e an Ha ies
3=BW L as well as dead ise and la e angle. The adius 2hal es he angle be ween 1and 3and is
chosen such ha he leng h p0-p1-p2-p3-p4is educed. The su ace be ween p3and p4is ex ended linea ely
up o a u he chine well abo e he wa e line o a oid he in oduc ion o an addi ional chine a he wa e line.
A hull gene a ed using he ou pla e model is shown in Fig. 5. The wa e line is ma ked in ed.
Figu e 5: Full hull using he ou pla e model
The models comp ise mo e han 20 pa ame e s comple ely de ining he hull’s shape. Se en o eigh
o hese pa ame e s (depending on he model h ee s. ou pla es) ha e been sys ema ically a ied while
he emaining pa ame e s we e ixed o meaning ul alues de e mined in a p elimina y e alua ion phase.
Compa ed o con en ional spline-based app oaches using an uncon ollable la ge numbe o e ices he
numbe o pa ame e s is e y low, bu s ill allows o a wide ange o possible hull shapes. I should be
men ioned ha no all pa ame e combina ions lead o alid hull shapes: Ou o 6000 pa ame e combina ions
o he po en ial low analysis only a ound 4500 esul ed in a easable hull shapes. The numbe o easable
designs wi hin a pa ame e ange is an indica o o he s abili y o he geome y o a pa ame ic model. The
la ge he allowable ange o each pa ame e and he mo e ex eme he allowable pa ame e combina ions a e
he mo e lexible he model is. This is pa icula ly impo an i non-con en ional hull shapes a e o in e es .
I was ound ha he pa ame e s o bow ineness and s e n wid h showed highes sensi i i y owa ds he
op imiza ion goals. I also seems ha he alignmen o he ha d chines wi h he low a ound he hull s ongly
in luences he pe o mance. This does no come as a big su p ise. Howe e , an alignmen o he chines wi h
he low does no lend i sel o easily accessible design pa ame e s wi hou iola ing he ai ness o he lines.
The e o e, only wi h a sys ema ic sea ch he mos a ou able designs could be ound.
2.3 CFD Analysis
The main ad an age o a pa ame ic model c ea ed in CAESES®is i s uni e sal applicabili y o di e en
CFD codes such as SHIPFLOW®and FINE™/Ma ine. Fo SHIPFLOW®a s a ion based geome y expo
was used while FINE™/Ma ine equi es essela ions o he comple e calula ion domain in he o m o mul i-
body STL iles. These simula ions can be highly au oma ed. The simula ion imes a y gea ly om a ound
30 minu es o a wa e esis ance calcula ion using SHIPFLOW®’s po en ial low sol e , ia a ound 4 hou s
o SHIPFLOW®’s zonal app oach, bo h on a con en ional 3,5GHz 4 co e PC wi h 16Gb RAM, all he way
up o mo e han 30 hou s on a 3.0GHz 8 co e PC wi h 64Gb RAM o a iscous esis ance calcula ion wi h
a ee su ace o an appended wo-hull con igu a ion using FINE™/Ma ine.
Fo he selec ed designs hei o al esis ance o e he comple e ope a ion speed ange was calcula ed
o e i y ha he lowe esis ance a Fn=0.3 is no penalized a o he eloci ies. Fig. 6 shows ha he
esis ance hump a Fn=0.3 could be mos ly elimina ed wi hou sac i icing pe o mance a highe speeds.
The ou pla e model e en pe o ms be e a highe speeds han he e e ence design.
In able 1 he nume ical esul as well as some geome ical and hyd os a ic pa ame e s o he designs a e
summa ized. The main objec i e o educing he o al esis ance a Fn=0.3 was well achie ed. The bes
h ee pla e design shows a o al esis ance educed by 13.5%, he bes ou pla e design shows a educ ion o
4
327
Bodo Hasubek and S e an Ha ies
Figu e 6: To al Resi ance o he selec ed h ee and
ou pla e designs compa ed o he e e ence design
Figu e 7: To al esis ance o he ou pla e model
e alua ed wi h SHIPFLOW®and FINE™/Ma ine
17.1%. A he same ime he pe omance a Fn=0.44 did no su e and e en sligh ly dec eased as well o
he la e . Mo eo e , an in e es ing esul is ha beyond an L/B a io o 10 i is no gua an eed ha a la ge
L/B alue s ands o be e pe o mance. Figu es 8 and 9 show he wa e pa e ns o he bes h ee and ou
pla e designs compa ed o he e e ence design a Fn=0.3 (8.2kno s) and 0° heel. The imp o emen s in
he wa e pa e ns show a he o e pa o he hull and in he wa e elimina ion in he wake.
Pe o mance Pa ame e s Re e ence Design Th ee Pla e Design Fou Pla e Design
R (Fn=0.3)[N] 1723 1490 1428
R (Fn=0.44)[N] 5277 5274 5195
T im a Fn=0.44 [°] 0.82 0.83 0.86
Hull Pa ame e s Re e ence Design Th ee Pla e Design Fou Pla e Design
LWL [m] 20.1 20.1 20.17
BWL [m] 1.60 1.90 1.84
Tc[m] 1.05 1.02 0.97
We ed su ace [m2]46.7 46.1 46.4
L/B12.6 10.6 10.95
BTR 1.52 1.87 1.9
Cp0.634 0.588 0.595
Table 1: Nume ical CFD esul s and hull pa ame e s o he demi-hull
In o de o e i y he alidi y o he esul s, he esis ance cu e o he bes ou pla e design was e-
e alua ed wi h a di e en CFD code (NUMECA’s FINE™/Ma ine) which uses a ee-su ace RANSE ap-
p oach (Fig. 7). The maximum de ia ion be ween calcula ed alues lies a Fn=0.36 and eaches 5.5%,
while a he op imiza ion poin s Fn=0.3 and Fn=0.44 he de ia ion is 1.8% and 0.1%, espec i ely.
The e o e, he op imiza ion p ocedu e can be assumed o deli e eliable esul s wi hin an accep able ma gin
o e o o a c uising ca ama an. I is no ed ha he ela i e posi ioning o he di e en designs wi h ega d
o hei pe o mance is usually e en mo e eliable han hei absolu e esis ance alue i he compa a i e
e alua ions a e pe o med using he same mesh esolu ion and he same sol e (as was done he e).
5
328

Bodo Hasubek and S e an Ha ies
Figu e 8: Demi-hull a Fn=0.3: Re e ence design s. bes h ee pla e design a 0° heel
Figu e 9: Demi-hull a Fn=0.3: Re e ence design s. bes ou pla e design a 0° heel
6
329
Bodo Hasubek and S e an Ha ies
One hull Two hulls ca ama an con igu a ion
FnResis ance [N] BCB [m] Resis ance ∆R
0.3 2872 4.25 2908 1.3%
0.44 10568 4.00 12908 22.1%
0.44 10568 4.25 12506 18.3%
0.44 10568 4.50 12258 16.0%
Table 2: Resis ance o one hull s. wo hulls ca ama an con igu a ion
2.4 Closing he design loop
A u he ad an age o a pa ame ic model is he abili y o e-use all he p ope ies o he single hull canoe
body model and e ine i o include a double hull se up and appendages ( udde s and dagge boa ds). Such a
e ined model is used o de e mine he bes comp omise be ween he hull’s cen e lines dis ance (BCB) and
o al esis ance.
Table 2 shows he in luence o a single hull compa ed o a wo hull ca ama an con igu a ion. I can be seen
ha a Fn=0.3 he e is close o no in e e ence be ween he hulls, while a Fn=0.44 he hull in e ac ion
signi ican ly inc eases he o al esis ance. The wa e in e ac ion o Fn=0.44 is shown in igu es 10 and 11.
As expec ed, he da a show ha inc easing BCB educes he o al esis ance, bu he e ec slows down o
la ge BCBs. The e o e, a BCB o 4.25 me e s was aken as a comp omise.
Figu e 10: Wa e pa e n a Fn=0.44 one hull Figu e 11: Wa e pa e n a Fn=0.44 wo hulls
Fo a p elimina y pe o mance p edic ion o he ca ama an design he ully appended pa ame ic model is
used o gene a e he esis ance da a equi ed o a eloci y p edic ion p og am (VPP). Common VPP pack-
ages ely on hull da a ha is de i ed om he Del sys ema ic se ies wi h addi ions o heel and appendages.
While his wo ks sa is ac o ily o monohull designs wi h ounded sec ions, he da a basis o he Del se ies
does no suppo ha d chined mul ihull designs e y well. Fu he mo e, sailing ca ama ans wi h dagge -
boa ds a e no mally sailed wi h he leewa d dagge bo d down only. This esul s in a highly asymme ic
se up ha is no easily accessible o analy ical app oaches.
This p oblem is o e come by calcula ing he esis ance in he di ec ion o mo emen and he side o ces
gene a ed by he hull and appendages o heel angles om 0° o 5°, leeway angles be ween 0° and 4° and
o a eloci y ange be ween 4 and 14 kno s. The limi s we e chosen in acco dance wi h he an icipa ed
ope a ional p o ile. Fo example, he maximum heel was limi ed 5° since his ep esen s he angle a which
hal he maximum s a ic s abili y is eached. A c uising ca ama an should no be sailed beyond his poin o
sa e y easons.
Wi h ou poin s in each o hese dimensions his app oach leads o 64 calcula ions in o al. Fig. 12
7
330
Bodo Hasubek and S e an Ha ies
Figu e 12: Wa e igh iangula ion o he appended model
shows he wa e igh iangula ion o he asymme ically appended pa ame ic hull model. The asymme y
equi es a comple e calcula ion domain o a ound 6 million cells. The calcula ions we e pe o med on said
8 co e PC. They ook 77 days. Ob iously his ime ame needs o be educed o e e y day use. Howe e ,
high pe o mance compu e clus e s wi h all he so wa e equi ed p e-ins alled a e a ailable o en and can
pe om such calcula ions wi hin a ew days [5].
The esul ing o ces o he CFD calcula ions depend on he h ee independen a iables heel angle, leeway
angle and eloci y. Fig. 13 and 14 show an example o he esis ance and side o ces o a ixed heel angle
o 2.5°. Fo u he p ocessing in a VPP hese alues a e in e pola ed using a spline based sca e ed da a
in e pola ion [2]. In p e ious wo k [3] Hazen’s his o ic sail model is used o calcula e he equilib ium
o he hyd odynamic and he ae odynamic o ces o de i e he sailing pe o mance unde di e en sailing
condi ions. In he cu en case a mo e accu a e sail model based on he ORC VPP ae odynamic model [4]
is employed. Based on he CFD da a and he ae odynamic model he expec ed sailing pe o mance can be
p edic ed accu a ely in an ea ly design s age.
3 Simula ion-D i en Design o a as planing mo o boa
3.1 Design Objec i e
A small ha d-chine mo o boa o a ound 6 m leng h and 1.3m3displacemen , ypically used as a day
c uise on la ge lakes and as a ende o supe yach s in coas al a eas, was s udied wi h ega d o i s pe o -
mance a a design speed o 18 kn. Wi h a F oude numbe o 1.32, based on he leng h o he design wa e line
a es , his ep esen s a planing hull o which dynamic im and sinkage (o a he li ) and he in luence o
he ee su ace a e c ucial.
8
331
Bodo Hasubek and S e an Ha ies
Figu e 13: Resis ance o ces a heel =2.5° Figu e 14: Side o ces a heel =2.5°
Fo his ype o boa s model es campaigns a e o en p ohibi i ely expensi e. A s anda d app oach
he e o e is o ely on se ies da a ound in li e a u e, see [6, 7]. Once he main pa icula s a e se , he
in luence o small changes o he be e o wo se canno be de i ed om he se ies da a. Al e na i ely, a
sui able pa en hull is aken, sligh ly modi ied and subsequen ly buil – na u ally, hoping o easonably good
pe o mance.
Bo h design app oaches canno p o ide answe s o he ypical design ques ions o how dead ise, ocke ,
hollowness o he planing su ace, i s wa p owa ds he ansom, wid h and shape o he sp ay ail and
o he design pa ame e s should be chosen. In his si ua ion an ex ended in es iga ion based on simula ion-
d i en design can help. To his e ec , he geome y o he mo o boa was pa ame ically modeled wi hin
CAESES®while he low was compu ed wi h he high- ideli y code FINE™/Ma ine, coupled o CAESES®
o be au oma ically execu ed in a design loop.
3.2 Pa ame ic Model
As i is common in design wo k, an a ac i e and ele an baseline was chosen as a good s a ing poin .
He e he Ri a Junio , a mo o boa o classic beau y om he mid 1960s, was selec ed. A ully-pa ame ic
model was de eloped wi hin CAESES®[8]. Fo he pu pose o he s udy he pa ame ic model was buil o
suppo wide changes o he hull while being able o closely app oxima e he baseline.
Figu e 15: Building pa e n o he gene ic sec ion (le : sec ion in he a body, igh : sec ion in he o ebody
9
332