![• Examples!of!NWB!applica2ons!
• SWENSE(method:(poten8al(–(Navier(Stokes(coupling(
! Linear(and(non(linear(waves(propaga8on:(
• Airy(linear(regular(waves(
• Rienecker.Fenton(non(linear(regular(waves(
• JONSWAP(spectrum(for(irregular(waves(
• Cross(wave(genera8on(by(spectrum(superposi8on(
(
(
une m´ethode de type laplacien. La d´etection du d´eferlement est activ´ee, de fa¸con `a observer si elle
e sur ce cas qui normalement n’en pr´esente pas ´etant donn´e la lin´earit´e de la houle et l’absence
ans le bassin.
Param`etre Valeur
Ordre espace 3
Ordre temporel 1
Tol. r´esolution GMRES 1e 6
Remplissage ILU(n) n=1
Acceleration parall`ele « Robust »
CFL 0.6
Donn´ees de calcul : param`etres num´eriques utilis´es pour la simulation de houle lin´eaire en bassin.
s
s obtenus pour le calcul des hauteurs de surface libre, de vitesse et de pression sont identiques,
oit la m´ethode utilis´ee. Par ailleurs, nous avons aussi v´erifi´e la coh´erence des longueurs d’onde
alcul´ees avec les deux m´ethodes, avec encore de bons r´esultats, et sans di↵´erences entre les deux
Cette premi`ere validation bas´ee sur la comparaison des r´esultats obtenus avec deux m´ethodes de
ule di↵´erentes est donc satisfaisante. La figure 5 montre un clich´e de la houle au cours du calcul.
Hauteur de surface libre pour une houle r´eguli`ere et lin´eaire g´en´er´ee en bassin par la m´ethode
e Rienecker et Fenton.
e calcul et premiers tests de scalabilit´e
e temps physique, les r´esultats sont fournis dans le tableau 3. Les r´esultats de scalabilit´e sont
cres, il faudrait en revanche tester ces mˆemes calculs sur le cluster de LEMMA et non sur un
eur 4 coeurs comme celui qui a ´et´e utilis´e pour ces tests. Les r´esultats de scalabilit´e seraient sans
meilleurs.
Param`etre Valeur [SI]
Diam`etre 0.625
Longueur immerg´ee 0.9375
Facteur d’´echelle 1
Table 5 – Caract´eristiques g´eom´etriques du cylindre tronqu´e.
Figure 6 – Houle r´eguli`ere non lin´eaire g´en´er´ee en bassin par la m´ethode de Rienecker et Fenton.
5.2 Cylindre tronqu´e soumis `a une houle r´eguli`ere
5.2.1 Description du cas
Les caract´eristiques g´eom´etriques de ce cylindre immerg´e sont fournies dans le tableau 5 ci-dessous. Cette
g´eom´etrie a fait l’objet d’une campagne exp´erimentale r´ealis´ee `a MARINTEK par Krokstad et Stanberg [6].
La houle g´en´er´ee est r´eguli`ere, et poss`ede les caract´eristiques pr´esent´ees sur le tableau 6. Si l’on applique le
regular(non(linear(regular(linear( JONSWAP(cross(waves(
16(
Wave!and!wave!interac2on!modeling!
this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!](https://image.slidesharecdn.com/presentationanamar-150610124732-lva1-app6891/85/Presentation-Numerical-Wave-16-320.jpg)
![• VIM!of!buoys!
(
(
(
(
• VIM!of!plaborms!
(
(
)
s
V
n
w
d
d
J
n
n
r
e
g,
D
d
r.
y
s
n
d
g
st
g
a)
b)
Figure 5: Flow visualization over the Slender Buoy
Note: a) Visualisation by 2D slice of the iso-contours of the
velocity magnitude
b) Visualization by iso-surface of the pressure
Focusing on 0° current heading, the well-known ‘bell-shape’ of
the VIM response curve with regard to the reduced velocity is
reminiscent to results obtained in [2] at lower Re, refer to
Figure 6. With increasing current speed from 0.15m/s up to
0.3m/s, the VIM amplitude increases for the In-Line, Cross-
Flow and Yaw DOF and then decreases from 0.3m/s up to
0.68m/s. At higher speed (1.31m/s) the VIM and VIR
amplitudes increase. Such phenomenon is more pronounced for
the yaw since the lock-in seems initiated at Vr≈4.
ICARE globally predicts larger recirculation at the leading
sharp edge. The difference observed in flow structures could be
explained by the use of several turbulence models in the three
solvers. Such behavior may be highlighted later in the paper
when dealing with hydrodynamic properties of the buoy.
Figure 6 – Flow features visualizations obtained by ICARE,
EOLE and ANANAS respectively from left to right; Iso-
contour of the velocity modulus over the Flat buoy tilted of
6° (1.3m/s corresponding full-scale current velocity).
Moreover, an associated period can be defined for those
vortical structures recovered by CFD analyses. Associated
power spectra appear less rich than those resulting from wind
tunnel tests but exhibit flow frequency excitation as previously
Vivet'et'al.'OMAE'2011'Minguez'et'al.'OMAE'2011'
slender(buoy( bundle( flat(buoy(
Sirnivas'et'al.'OMAE'2006'
Deep(Chamber(Column(
courtesy(of(INTECSEA(
Straked(cylinder(and(SPAR(
19(
courtesy(of(TECHNIP(
courtesy(of(TECHNIP(
courtesy(of(TECHNIP( courtesy(of(TECHNIP(
Turbulent!flows!
this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!](https://image.slidesharecdn.com/presentationanamar-150610124732-lva1-app6891/85/Presentation-Numerical-Wave-19-320.jpg)
Presentation Numerical Wave
- 2. Outline!of!the!presenta2on! 1. Introduc2on!to!the!company! . General(presenta8on(of(the(company( . Original(methods(and(technologies( 2. Introducing!ANAMAR!the!new!NWB!standGalone!solver! . General(presenta8on(of(the(NWB( . Original(methods(and(technologies( ! 3. Our!offer! 2( this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 3. Paris((2012)( Houston((2011)( Toulouse((2013)( • R&D!and!soMware!company!created!in!2001! • Staff:! . 20(engineers(including(12(PhDs( . 4(PhD(students((INRIA/LEMMA)(((( • 4!loca2ons!!! Who!are!we?! Sophia.An8polis( 3( this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 6. • High!performance!compu2ng!and!visualisa2on! • Latest(genera8on(cluster((IvyBridge(–(SandyBridge)( 600!compu2ng!cores!.>(up(to(200Mnodes! • ((Parallel(visualisa8on(work(sta8ons( • !InGhouse!soMwares!for!preGprocessing!and!solver! • (Meshing(and(modeling( • Unstructured(and(adap8ve(mesh( • ((Fluid(and(Structure(Dynamics(solver( • Monolithic(algorithm( • Mul8physics(solver( • (Modal(sloshing(analysis(( Our!tools! 6( this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 8. ( MESH! (( ( PHYSICS! (( ( NUMERICS! (( Small and large structure deformation Free surface flows Multiphase flows (In)compressible flowsTurbulence (RANS/VMS) Non newtonian flows Oil / polymer rheology High order conservative schemes at high anisotropy Mixed Finite Element and Volume methods Heat transfer fluid and structure Emulsion / sand flow Fluid - Structure Interaction Contact / damaging Cryogenic fluids 8( Our!technologies! Up to 6th order in space and 3rd order in time 2D/3D Fully unstructured and intelligent mesh Moving mesh technology this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 10. • Illustra2on!of!mul2Gphysics!problems!handled!by!ANANAS! • Fluid(Structure(Interac8on:( • Large(displacement( • Complex(geometry( ( • Ice(simula8ons:( • Three.phase(flow( • Contact(/(damaging( ( • Viscoplas8c(flow(in(pumps:( • FSI(and(non.newtonian(fluids( courtesy(of(PCM( courtesy(of(DECATHLON( 10( Mul2Gphysics!monolithic!solver! ice(loads(on(conical(plaaorm( flipper( Progressive(Cavity(Pump( this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 12. • Numerical!Wave!Basin!–!general!descrip2on! ! ( . A(simula8on(tool(to(study(sea(.(keeping(and(maneuverability( . Complex(body(geometry((propeller,(rudder,(appendages)(and(mo8on((6(DOF)( . Metocean(condi8ons((waves,(wind,(current,(bathymetry)( . Mooring((linear,(non.linear)( ( WIND( 12( ANAMAR!–!stand!alone!NWB! WAVES( CURRENT( damping( The grid used to solve SWENSE equations for the diffracted flow does not need to be adapted to the description of the incident wave field, which is given explicitly in the whole water and air domain by the nonlinear potential flow model. Thus, it enables us to realize simulations 50 times faster than usual direct methods, without any W m this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 13. • Numerical!Wave!Basin!featuring:! • Free(surface(modeling(with(Level.set(method( • Wave(making( ! Direct(Navier(Stokes(resolu8on(of(the(waves(propagated(from(a(wave( maker(in(order(to(reproduce(basin(condi8ons( ! Regular(and(irregular(wave(condi8ons(with(a(fast(Navier.Stokes(poten8al( coupling(method( • Wind(and(current( ! Viscid(and(inviscid(flows( ! Reynolds(Averaged(Navier.Stokes((RANS)(and(Large(Eddy(Simula8on((LES)( for(turbulent(flows( • Fluid(Structure(Interac8on(for(rigid(and(elas8c(bodies( • Coupling(with(ANAMESH(for(automa8c(refinement( 13( General!features! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 14. • Direct!NavierGStokes!approach!for!wave!propaga2on! ! Regular(and(irregular(waves(can(be(taken(into(account(by(applying(the( wave(maker(velocity(to(the(corresponding(wall(of(the(numerical(wave(basin( to(mimick(the(wave(maker.( ! The(free(surface(is(taken(into(account(with(the(Level.set(method((up(to(3rd( order(in(space.( ! Air(can(either(be(taken(into(account(or(neglected.( ( ( ( regular(non(linear( irregular( 14( Wave!modeling! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 15. • SWENSE!method:!poten2al!–!Navier!Stokes!coupling! ( ! The(SWENSE(approach(is(based(on(a(physical(decomposi8on(by(combining( incident(waves(described(by(a(nonlinear(spectral(scheme(based(on(poten8al( flow(theory(and(a(Navier.Stokes(solver(where(only(the(diffracted(part(of(the( flow(is(solved,(incident(flow(parameters(seen(as(forcing(terms,(damping(is( applied(in(the(far(field(on(the(diffracted(terms.( ( 15( Luquet'PhD'thesis'2005' Wave!and!wave!interac2on!modeling! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 16. • Examples!of!NWB!applica2ons! • SWENSE(method:(poten8al(–(Navier(Stokes(coupling( ! Linear(and(non(linear(waves(propaga8on:( • Airy(linear(regular(waves( • Rienecker.Fenton(non(linear(regular(waves( • JONSWAP(spectrum(for(irregular(waves( • Cross(wave(genera8on(by(spectrum(superposi8on( ( ( une m´ethode de type laplacien. La d´etection du d´eferlement est activ´ee, de fa¸con `a observer si elle e sur ce cas qui normalement n’en pr´esente pas ´etant donn´e la lin´earit´e de la houle et l’absence ans le bassin. Param`etre Valeur Ordre espace 3 Ordre temporel 1 Tol. r´esolution GMRES 1e 6 Remplissage ILU(n) n=1 Acceleration parall`ele « Robust » CFL 0.6 Donn´ees de calcul : param`etres num´eriques utilis´es pour la simulation de houle lin´eaire en bassin. s s obtenus pour le calcul des hauteurs de surface libre, de vitesse et de pression sont identiques, oit la m´ethode utilis´ee. Par ailleurs, nous avons aussi v´erifi´e la coh´erence des longueurs d’onde alcul´ees avec les deux m´ethodes, avec encore de bons r´esultats, et sans di↵´erences entre les deux Cette premi`ere validation bas´ee sur la comparaison des r´esultats obtenus avec deux m´ethodes de ule di↵´erentes est donc satisfaisante. La figure 5 montre un clich´e de la houle au cours du calcul. Hauteur de surface libre pour une houle r´eguli`ere et lin´eaire g´en´er´ee en bassin par la m´ethode e Rienecker et Fenton. e calcul et premiers tests de scalabilit´e e temps physique, les r´esultats sont fournis dans le tableau 3. Les r´esultats de scalabilit´e sont cres, il faudrait en revanche tester ces mˆemes calculs sur le cluster de LEMMA et non sur un eur 4 coeurs comme celui qui a ´et´e utilis´e pour ces tests. Les r´esultats de scalabilit´e seraient sans meilleurs. Param`etre Valeur [SI] Diam`etre 0.625 Longueur immerg´ee 0.9375 Facteur d’´echelle 1 Table 5 – Caract´eristiques g´eom´etriques du cylindre tronqu´e. Figure 6 – Houle r´eguli`ere non lin´eaire g´en´er´ee en bassin par la m´ethode de Rienecker et Fenton. 5.2 Cylindre tronqu´e soumis `a une houle r´eguli`ere 5.2.1 Description du cas Les caract´eristiques g´eom´etriques de ce cylindre immerg´e sont fournies dans le tableau 5 ci-dessous. Cette g´eom´etrie a fait l’objet d’une campagne exp´erimentale r´ealis´ee `a MARINTEK par Krokstad et Stanberg [6]. La houle g´en´er´ee est r´eguli`ere, et poss`ede les caract´eristiques pr´esent´ees sur le tableau 6. Si l’on applique le regular(non(linear(regular(linear( JONSWAP(cross(waves( 16( Wave!and!wave!interac2on!modeling! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 17. • Coupling!with!mesh!deforma2on!or!intelligent!mesh!! ! Arbitrary(Lagrangian(Eulerian((ALE)(approach(for(free(surface(tracking:(saves( points(in(the(resolu8on(of(wave(but(limited(to(non(breaking(waves( ( ! Intelligent(mesh(for(anisotropic(mesh(adapta8on(around(the(free(surface(and( boundary(layer(genera8on(( cylinder(run.up( mesh(of(a(skiff(geometry(with(boundary(layer(NACA(24(Fr(0.37( 17( FreeGsurface!modeling! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 18. • Modeling!of!current!around!floa2ng!structures! ! Surface(piercing(structures(with(current(and(waves( ! Flat(/(profiled(current,(thermally(stra8fied(flows( ! RANS(models((k.epsilon,(Spalart.Allmaras)(for(very(high(Reynolds(flows( ! Scale(resolving(methods((LES,(LES.Varia8onal(Mul8scale(.(VMS(.(hybrid( RANS.VMS,(Detached(Eddy(Simula8on(–(DES):(moderate(to(high(Reynolds( ( ( Re(number( NACA(FR(0.37(RANS(Full(Spar(VMS((Tandem(cylinders(VMS( Sainte6Rose'et'al.'OMAE'2014' Sirnivas'et'al.'OMAE'2006' Square(cylinder(VMS( 10,000( 100,000( 1M( 18( Turbulent!flows! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 19. • VIM!of!buoys! ( ( ( ( • VIM!of!plaborms! ( ( ) s V n w d d J n n r e g, D d r. y s n d g st g a) b) Figure 5: Flow visualization over the Slender Buoy Note: a) Visualisation by 2D slice of the iso-contours of the velocity magnitude b) Visualization by iso-surface of the pressure Focusing on 0° current heading, the well-known ‘bell-shape’ of the VIM response curve with regard to the reduced velocity is reminiscent to results obtained in [2] at lower Re, refer to Figure 6. With increasing current speed from 0.15m/s up to 0.3m/s, the VIM amplitude increases for the In-Line, Cross- Flow and Yaw DOF and then decreases from 0.3m/s up to 0.68m/s. At higher speed (1.31m/s) the VIM and VIR amplitudes increase. Such phenomenon is more pronounced for the yaw since the lock-in seems initiated at Vr≈4. ICARE globally predicts larger recirculation at the leading sharp edge. The difference observed in flow structures could be explained by the use of several turbulence models in the three solvers. Such behavior may be highlighted later in the paper when dealing with hydrodynamic properties of the buoy. Figure 6 – Flow features visualizations obtained by ICARE, EOLE and ANANAS respectively from left to right; Iso- contour of the velocity modulus over the Flat buoy tilted of 6° (1.3m/s corresponding full-scale current velocity). Moreover, an associated period can be defined for those vortical structures recovered by CFD analyses. Associated power spectra appear less rich than those resulting from wind tunnel tests but exhibit flow frequency excitation as previously Vivet'et'al.'OMAE'2011'Minguez'et'al.'OMAE'2011' slender(buoy( bundle( flat(buoy( Sirnivas'et'al.'OMAE'2006' Deep(Chamber(Column( courtesy(of(INTECSEA( Straked(cylinder(and(SPAR( 19( courtesy(of(TECHNIP( courtesy(of(TECHNIP( courtesy(of(TECHNIP( courtesy(of(TECHNIP( Turbulent!flows! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 22. • For!preGprocessing( ! boundary(condi8ons( ! hydrosta8c(hull(stabiliza8on( ! current(and(/(or(wind( ! choice(of(turbulence(model( ! linear(and(non.linear(moorings(( ! DOF( ! mechanical(system(including(mass(and(iner8a(matrix( ! interface(resolu8on(method((Navier.Stokes(or(SWENSE)( ! wave(proper8es((Airy,(Rienecker.Fenton,(JONSWAP)( ! wave(maker( 22( A!dedicated!Graphical!User!Interface! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 23. • Strong!R&D!staff!and!partnerships! ! Highly(qualified(applica8on,(support(and(sohware(engineers! ! Scien8fic(commiiee(coming(from(the(applied(mathema8cs( community(in(french(laboratories( ! Historical(partnership(with(INRIA(and(partnerships(with( universi8es(and(laboratories(in(Europe(and(USA( ! Partnership(on(numerical(problems(with(TOTAL!since(2010( ! More(than(30(publica8ons(in(the(past(two(years( ! Par8cipa8on(in(PhD(–(ready(projects(and(JIPs( • currently(hiring(for(a(JIP(on(«(Mul8.scale(Modeling(of(Erosion(»,(so(contacts( needed!( 23( Our!added!value!in!a!compe22ve!context! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 24. • SoMware!licensing! ! Affordable(cost(for(HPC(systems( ! Dedicated(support( ! On.demand(model(defini8on( ( ( ( ( ( ( ( Preprocessing ANAMESH® 2D/3D Mesh files, CAD (igs, stp,…) Postprocessing and analysis Paraview® / Tecplot® Solvers ANANAS® / ANASLOSH® / ANAMAR® Sequential or parallel User functions and files (physical laws, mesh displacement and metrics, source terms,…) 24( Our!added!value!in!a!compe22ve!context! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 25. • The! general! quality! process! to! monitor! the! soMware! development! and! valida2ons! complies! with! the! following! ECCS!standards!(ESA!standards):! – ECCS.Q.ST(for(product(assurance(aspects( – ECCS.E.ST( for( product( development( aspects( and( sohware( maintenance((ST.40C)(and(study(realiza8on( • Therefore! the! soMware! has! to! go! through! 4! levels! of! valida2on:! – Comparison(to(analy8cal(cases((required)!! – Comparison(to(experiments((desirable)( – Self.comparison(no.regression((required)! – Comparison(to(another(code((desirable)( • Consequently! more! than! 500! computa2ons! are! required! before!releasing!a!produc2on!version.! ( 25( Quality!assurance! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!
- 26. • A( verifica8on( (consistency,( convergence)( is( run( on( a( set( of( analy8cal( cases( and( simple( cases.( This( also( allows( to( get( numerical(recipes(for(similar(cases.( • For( the( grid( dependent( models( where( only( a( sta8s8cal( convergence(can(be(obtained((such(as(LES.VMS)(the(consistency( is(verified.( 26( SoMware!Verifica2on!&!Valida2on! this!document!is!confiden2al,!belongs!to!LEMMA!and!cannot!be!disclosed!to!any!third!party!without!wri<en!consent!