scieee Open visual document viewer

An iterative procedure to solve a coupled two-fluids turbulence model

Chacón Rebollo, Tomás; Pino, Stéphane del; Yakoubi, Driss

Abstract

This paper introduces a scheme for the numerical approximation of a model for two turbulent flows with coupling at an interface. We consider the variational formulation of the coupled model, where the turbulent kinetic energy equation is formulated by transposition. We prove the convergence of the approximation to this formulation for 3D flows for large turbulent viscosities and smooth enough flows, whenever bounded in W1,p Sobolev norms for p large enough. Under the same assumptions, we show that the limit is a solution of the initial problem. Finally, we give some numerical experiments to enlighten the theoretical work.

Full text

ESAIM: M2AN 44 (2010) 693–713 ESAIM: Ma hema ical Modelling and Nume ical Analysis DOI: 10.1051/m2an/2010015 www.esaim-m2an.o g AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL ∗ Tomas Chac´ on Rebollo1,S ´ ephane Del Pino2and D iss Yakoubi3 Abs ac . This pape in oduces a scheme o he nume ical app oxima ion o a model o wo u bu- len lows wi h coupling a an in e ace. We conside he a ia ional o mula ion o he coupled model, whe e he u bulen kine ic ene gy equa ion is o mula ed by ansposi ion. We p o e he con e gence o he app oxima ion o his o mula ion o 3D lows o la ge u bulen iscosi ies and smoo h enough lows, whene e bounded in W1,p Sobole no ms o pla ge enough. Unde he same assump ions, we show ha he limi is a solu ion o he ini ial p oblem. Finally, we gi e some nume ical expe imen s o enligh en he heo e ical wo k. Ma hema ics Subjec Classifica ion. 63N30, 76M10. Recei ed Ma ch 29, 2008. Re ised Augus 1s , 2009. Published online Feb ua y 23, 2010. 1. In oduc ion In his con ibu ion we ocus ou a en ion on he modelling o he su ace laye be ween he a mosphe e and he ocean. We a e in e es ed in designing effec i e p ocedu es o sol e he ollowing coupled model: ⎧ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎩ −∇·(αi(ki)∇ui)+g ad pi= iin Ωi, ∇·ui=0inΩ i, −∇·(γi(ki)∇ki)=αi(ki)|∇ui|2in Ωi, ui=0on Γi, ki=0onΓ i, αi(ki)∂niui−pini+κ(ui−uj)|ui−uj|=0on Γ,1≤i=j≤2, ki=λ|u1−u2|2on Γ. (1.1) Whe eeach iple(ui,p i,k i) is defined in he domain Ωi,1≥i≥2. The gene ic poin in R2, esp. inR3,is deno ed by x=(x, z), esp. x=(x, y, z). Keywo ds and ph ases. Ocean-a mosphe e coupling, u bulen lows, con e gence analysis, i e a i e me hod, spec al me hod. ∗T. Chac´on Rebollo was pa ially unded by EU Ma ie-Cu ie Fellowship P og amme, by Spanish Go e nmen G an MTM2006- 01275 and by Jun a de Andaluc´ıa G an P07-FQM-02538. 1Depa amen o de Ecuaciones Di e enciales y Analisis Nume ico, Uni e sidad de Se illa, Spain. 2CEA, DAM, DIF, 91297 A pajon, F ance. 3Labo a oi e Jacques-Louis Lions, Uni e si ´e Pie e e Ma ie Cu ie, 4 place Jussieu, 75005 Pa is Cedex, F ance. A icle published by EDP Sciences c EDP Sciences, SMAI 2010 694 T. CHAC ´ ON REBOLLO ET AL. Sys em (1.1) is a simplified model o wo s a iona y u bulen flows in adjacen domains, coupled by bounda y condi ions on he in e ace, such as he sys em a mosphe e-ocean. Indeed, i is a simplified ma hema ical o mula ion o he RANS (Reynolds A e aged Na ie -S okes) model o o de 1 used o simula e a s a iona y mean flow when con ec ion is neglec ed. This kind o modelling is o en used in enginee ing o geophysics, see o ins ance Be na di e al. [6], Launde and Spalding [16], Mohammadi and Pi onneau [21], Pique [22], Wilcox [23]. In wha ollows, Ωi(i=1,2) a e bounded domains o Rd,d=2,3, which a e ei he con ex o o class C1,1, wi h bounda ies ∂Ωi=Γ i∪Γ,Γ=Ω1∩Ω1being he in e ace be ween he wo fluids. Γ is assumed o be fla . Indeed, we assume ha he so-called “ igid lid hypo hesis” (in oduced by B yan in [9]) holds, an hypo hesis which is s anda d in geophysics and oceanog aphy. Each o he wo u bulen fluids is modeled by a simplified one-equa ion u bulence model whose unknowns a e he eloci y uiand he u bulen kine ic ene gy (TKE) ki. In he fi s equa ion we model he gene a ion o eddy iscosi y in flow iby he e m −αi(ki)∇ui. The (posi i e) quan i y αi(ki) is he eddy iscosi y. This is a simplifica ion o he usual modelling o Reynolds S ess Tenso by Ri≃−αi(ki)∇ui+∇ ui. We p e e he fi s exp ession o simplici y o ma hema ical analysis, al hough ou analysis s ill holds o he second one. We also neglec anspo effec s, we in end o analyze hem in a o hcoming pape . The fluids a e assumed o be incomp essible (second equa ion). In he hi d equa ion we model he gene a ion o TKE by means o a p oduc ion sou ce e m αi(ki)|∇ui|2, al hough he physical one should be αi(ki)|∇ui+∇ ui|2. Again, we p e e he fi s exp ession o simplici y o ma hema ical ea men . Also he u bulen diffusion o TKE is he unc ion γi(ki). We neglec he iscous dissipa ion effec s, o a oid o manage an addi ional s a is ic o he u bulence (a mixing leng h o he u bulen dissipa ion ε, o ins ance). We assume non- slipping bounda y condi ions in he bounda y pa s Γi o simplici y ( ou h and fi h equa ions). These in p ac ice a e eplaced by wall-laws o simula e he gene a ion o u bulence on solid bounda ies. The six h equa ion globally models he in e ac ion o he wo bounda y laye s on one and ano he side o he in e ace Γ as ic ion effec s, by means o a se o bounda y condi ions simila o Manning’s law. Finally, he las equa ion models he p oduc ion o TKE in he in e ace. The coefficien s κiand λa e posi i e. We assume ha he u bulen diffusions αiand γibelong o W1,∞(R)and e i yαi≥ν, γi≥ν, o some ν>0. The eddy diffusions usually a e unbounded unc ions o he TKE o he o m a+b√k,asweusein he nume ical simula ions epo ed in Sec ion 5(see o ins ance [8,17,21]). Bu his ende s he analysis much mo e complex e en o a one-fluid u bulence model (see [18]). So we conside a simplified model, ha s ill includes se e al ealis ic non-linea in e ac ions. Sys em (1.1) was s udied in [3] whe e exis ence and uniqueness o small smoo h solu ions we e p o ed. Spec al and Fini e Elemen disc e iza ions we e s udied in subsequen pape s by he same au ho s and co- wo ke s (see [4,5]). In hese pape s, he abili y o hese disc e iza ion echniques o app oach he solu ion o model (1.1) was p o ed. Howe e , in bo h cases he disc e iza ions achie ed consis ed in ully non-linea se s o algeb aic equa ions. Ou pu pose he e is o de i e i e a i e p ocedu es o sol e sys em (1.1) ha decouples he in e ac ion o he p oblem, leading o mildly non-linea p oblems. Le us in oduce he unc ion spaces Xi= i∈H1(Ωi); i=0on Γi, L2 0(Ωi)=qi∈L2(Ωi); Ωi qi=0 .(1.2) AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL 695 Conside also wo conjuga e posi i e eal numbe s and i.e. 1 +1 =1,such ha >d. We in oduce he ollowing i e a i e p ocedu e o sol e (1.1): once known un i∈Xi,pn i∈L2 0(Ωi), ki∈W1, (Ωi), i=1,2, sol e: P oblem 1. Ob ain un+1 i∈Xi,i=1,2, such ha Ωi αi(kn i)∇un+1 i:∇ idx−Ωi (∇· i)pn+1 idx+κiΓ|un+1 i−un+1 j|(un+1 i−un+1 j)· idτ=Ωi i· idτ, ∀ i∈Xi,and Ωi (∇·un+1 i)qidx=0,∀qi∈L2 0(Ωi), and P oblem 2. Ob ain kn+1 i∈W1, (Ωi),i=1,2, such ha kn+1 i=0 onΓ i,k n+1 i=λ|un+1 1−un+1 2|2on Γ, Ωi γi(kn i)kn+1 iϕidx=Ωi αi(kn i)|∇un+1 i|2ϕidx,∀ϕi∈W1, 0(Ωi). Rema k 1.1. We ake >d o gi e a sense o he equa ion o he kiin model (1.1). Indeed, he e m αi(kn i)|∇un+1 i|2belongs o L1(Ωi), and i ollows om he Sobole Imbedding Theo em ha he es unc ion ϕi in P oblem 2 belongs o L∞(Ωi), so ha he igh -hand membe is well defined. Obse e ha P oblem 1 is in eali y non-linea due o he p esence o he Manning-like sou ce e m. This is a mild non-linea i y due o he mono onic na u e o his e m, ha may be made explici in p ac ice i mass-lumping echniques a e used. Ou main esul s a es ha i he sequences (un i)nand (kn i)na e espec i ely bounded in W1,3+ε(Ωi)dand W1,3(Ωi), hen, o small enough da a (in a con enien sense), he i e a i e scheme is con ac ing. This egula i y is ealis ic, as i is no a om he W1,2 egula i y ha has been p o ed o p oblem (1.1) o gene al da a. The main ing edien s o show he con e gence o ou scheme a e he con enien choices o es unc ions, and he use o he ha monic li ings Rio Di ichle bounda y condi ions on Γ on he Ωi(see he p oo o Thm. 3.4). Ou pape is o ganised as ollows. In Sec ion 2we in oduce a weak o mula ion o he abo e i e a i e p ocedu e. Sec ion 3is de o ed o p o e he con ac i eness o he TKE sequence. Due o he p oduc ion e m o he TKE on in e ace Γ: ki=λ|u1−u2|2, i is necessa y o es ima e he exp ession |un+1 1−un+1 2|2−|un 1−un 2|2H 1 2 00(Γ) ,(1.3) whe e he special space H 1 2 00(Γ) is he subspace o H1 2(Γ) whose ex ension by ze o o ∂Ω1( o ins ance, i could be also o ∂Ω2) belongs o H1 2(∂Ω1). An in insic scala p oduc on H 1 2 00(Γ) is defined as ((u, ))H 1 2 00 (Γ) =Γ u(x) (x)dx+ΓΓ (u(x)−u(y)) ( (x)− (y)) |x−y|ddxdy+Γ u(x) (x) d(x, ∂Γ) ,(1.4) whe e he fi s wo summands define he H1 2(Γ) scala p oduc (see Adams and Fou nie [1], Thm. 7.48). I s exp ession in ol es he dis ance d(x, ∂Γ) o he bounda y o ∂Γ. I comes om he es ic ion o H 1 2 00(Γ) o he scala p oduc in H1 2(∂Ω1) o ins ance. I is gi en by Lions and Magenes in [20], Chap e 1, Theo em 11.7. 696 T. CHAC ´ ON REBOLLO ET AL. The es ima ion o (1.3) is done using a G is a d’s esul (Lem. 3.5 in he pape a hand, see [14] o he o iginal e e ence) on es ima es in Ws,p o p oduc s o unc ions o Wsj,pj, o some eal numbe s s, sjand some non-nega i e in ege s p, pj,j=1,2. Conce ning he sequence o p essu e i e a es (pn i)n, we use a specific in -sup condi ion (see Co . 3.7) o show ha i is a Cauchy sequence, see Theo em 3.8. The con e gence analysis is pe o med in Sec ion 4.InTheo em4.1 we p o e ha he iple (un i,kn i,p n i)n has a unique limi , which is a solu ion o he a ia ional o mula ion (2.3)–(2.4). We finally p esen some nume ical es s in Sec ion 5. These es s a e ealized wi h he so wa e F eeFEM3D (see [12]) in meaning ul si ua ions, ha ag ee wi h he expec a ions o ou esul . 2. I e a i e scheme We shall a fi s desc ibe he weak o mula ion o p oblem (1.1). We assume ha αiand γia e bounded unc ions om he se o nonnega i e eal numbe s R+on o R, and belong o W1,∞(Ωi), which sa is y ∀∈R+,δ 1≥αi()≥νand δ1≥γi()≥ν, (2.1) and ∀∈R+,|α i()|≤δ2and |γ i()|≤δ2,(2.2) whe e δ1,δ 2and νa e posi i e cons an s. Sys em (1.1) admi s he ollowing a ia ional o mula ion: Find (ui,p i,k i)∈Xi×L2(Ωi)×W1, (Ωi) such ha , o all ( i,q i,ϕ i)∈Xi×L2(Ωi)×W1, 0(Ωi), ai(ki;ui, i)+bi( i,p i)+κiΓ|ui−uj|(ui−uj)· idτ=Ωi i· idx bi(ui,q i)=0,(2.3) and, ki=0onΓ i,k i=λ|ui−uj|2on Γ,and Ci(ki;ki,ϕ i)=Ωi αi(ki)|∇ui|2ϕidx,(2.4) whe e he o ms ai(·;·,·),b i(·,·)andCi(·;·,·) a e defined by ai(i;ui, i)=Ωi α(i)∇ui:∇ idx, bi( i,q i)=−Ωi qi∇· idx, Ci(i;ki,ϕ )=Ωi γi(i)∇ki·∇ϕidx. No e ha he bilinea o ms aiand Ciin (2.3) depend on ki. Rema k 2.1. Since ui∈Xi hen i s ace on Γ belongs o H 1 2 00(Γ). Thus by using defini ion o his space ui|Γ belongs o H1 2(Γ) and applying he Sobole embedding om H1 2(Γ) in o L3(Γ)d, we conclude ha he in eg al Γ|ui−uj|(ui−uj)· idτis well defined. AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL 697 This o mula ion makes sense, as αi(ki)|∇ui|2∈W1, (Ωi)whenui∈Xi. In Lewandowski [19]i isp o ed ha his o mula ion admi s a leas a solu ion. We shall conside he ollowing i e a i e p ocedu e: Gi en (un i,p n i,kn i)∈Xi×L2(Ωi)×W1, (Ωi),i=1,2, ob ain (un+1 i,p n+1 i,kn+1 i)∈Xi×L2(Ωi)×W1, (Ωi), such ha ∀( i,q i,ϕ i)∈Xi×L2(Ωi)×W1, 0(Ωi), ai(kn i;un+1 i,∇ i)+bi( i,p n+1 i)+κiΓ|un+1 i−un+1 j|(un+1 i−un+1 j)· idτ=Ωi i· idτ, (2.5) and bi(un+1 i,q i)=0,(2.6) and kn+1 i=0onΓ i,(2.7) kn+1 i=λ|un+1 1−un+1 2|2on Γ,(2.8) Ci(kn i;kn+1 i,ϕ i)=Ωi αi(kn i)|∇un+1 i|2ϕidx.(2.9) 3. Con ac i eness In his sec ion we p o e ha he sequence o TKE (kn i)nis con ac ing, and ha consequen ly he sequences o eloci ies (un i)nalso is con ac ing, in he sense ha ⎧ ⎪ ⎪ ⎪ ⎪ ⎨ ⎪ ⎪ ⎪ ⎪ ⎩ 2  i=1 ∇(un+1 i−un i)2 L2(Ωi)≤K 2  i=1 ∇(kn i−kn−1 i)2 L2(Ωi),and 2  i=1 ∇(kn+1 i−kn i)2 L2(Ωi)≤K 2  i=1 ∇(kn i−kn−1 i)2 L2(Ωi). (3.1) We may in e p e hese inequali ies in he sense ha he sequence o pai s (un i,kn i)nis con ac ing in he Hilbe space Xi×L2(Ωi). Howe e , o simpli y ou de i a ion, we shall no explici ly use his space. Finally, we show ha he p essu es (pn i)nis a Cauchy sequence. We suppose om now on ha he sequences (un i)nand (kn i)n e i y he ollowing hypo hesis. Hypo hesis 3.1. ∀n∈N,un i∈W1,3+ε(Ωi)dand kn i∈W1,3(Ωi), and one has un iW1,3+ε(Ωi)d≤M, kn iW1,3(Ωi)≤M, whe e Mand εa e wo ixed posi i e numbe s. Rema k 3.2. No e ha he na u al es ima es o eloci ies in model (1.1)a einH1no m, no in W1,3no m. Indeed, choosing iequal o 1 κiun+1 i∈Xiin equa ion (2.5), and summing upon i=1,2gi es 2  i=1 1 κiΩi αi(kn i)|∇un+1 i|2dx+Γ|un+1 1−un+1 2|3dτ= 2  i=1 1 κiΩi iun+1 idx. Since he in eg a ed e m on Γ is nonnega i e and hanks o (2.1), we deduce ν cM 2  i=1 ∇un+1 i2 L2(Ωi)d≤1 cm 2  i=1 Ωi iun+1 idx. 698 T. CHAC ´ ON REBOLLO ET AL. Using he Cauchy-Schwa z and Poinca ´e-F ied ichs inequali ies, we ob ain 2  i=1 ∇un+1 i2 L2(Ωi)d≤c ν2 2  i=1  i2 L2(Ωi)d,(3.2) whe e cis a posi i e cons an , depending only on he domains Ωiand he ic ion coefficien s κi. We nex p o e ha he con ac i eness o he TKE implies ha o he eloci ies. Lemma 3.3. Assume ha Hypo hesis 3.1 holds and ha i∈L2(Ωi)d,i=1,2. Then he e exis s a posi i e cons an c, depending only on Ωi, such ha 2  i=1 ∇(un+1 i−un i)2 L2(Ωi)d≤cδ2 2M2 ν2 2  i=1 ∇(kn i−kn−1 i)2 L2(Ωi)d.(3.3) P oo . Le us ake i=1 κi (un+1 i−un i)∈Xias a es unc ion in (2.5)a i e a ionsnand n+ 1. Then, calcula ing he diffe ence be ween bo h ob ained equa ions, and summing on i=1,2, yields 2  i=1 1 κiΩi αi(kn)∇un+1 i−un i:∇un+1 i−un idx + 2  i=1 1 κiΩiαi(kn)−αikn−1 i∇un i:∇un+1 i−un idx +Γun+1 1−un+1 2un+1 1−un+1 2−|un 1−un 2|(un 1−un 2)·un+1 1−un+1 2−(un 1−un 2)dτ=0. The ollowing inequali y holds o all ec o s a,b∈Rd, (|b|b−|a|a)·(b−a)≥0.(3.4) To p o e i , conside he unc ion J:Rd→ Rdefined by J(a)=2 3|a|3.Jis con ex and diffe en iable. Thus, (∇J(a)−∇J(b)) ·(b−a)≥0,∀a,b∈Rd. Then, (3.4) ollows as ∇J(a)=|a|a·b. We deduce ha 2  i=1 1 κiΩi αi(kn i)|∇(un+1 i−un i)|2dx+ 2  i=1 1 κiΩi (αi(kn i)−αi(kn−1 i))∇un i·∇(un+1 i−un i)dx≤0.(3.5) AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL 699 I comes om Hypo hesis 3.1 ha ∇un ibelongs o L3(Ωi)dand ha ∇un iL3(Ωi)d≤M. Fu he mo e, acco ding o he ela ion (2.2) and he canonical injec ion om H1(Ωi) oL6(Ωi)andH¨olde inequali y, we ob ain ν 2  i=1 ∇(un+1 i−un i)2 L2(Ωi)d≤δ2 2  i=1 Ωi|kn i−kn−1 i||∇un i||∇(un+1 i−un i)|dx ≤δ2 2  i=1 kn i−kn−1 iL6(Ωi)∇un iL3(Ωi)d∇(un+1 i−un i)L2(Ωi)d, ≤δ2 2M2 2ν 2  i=1 kn i−kn−1 i2 L6(Ωi)+ν 2 2  i=1 ∇(un+1 i−un i)2 L2(Ωi)d. F om his es ima e we conclude ela ion (3.3).  We nex p o e he con ac i eness o he sequence o TKE (kn i)n. Theo em 3.4. Assume ha Hypo hesis 3.1 holds and ha i∈L2(Ωi)d. Then he e exis s a posi i e cons an c, depending only on Ωiand on he da a κiand λ, such ha o all n∈N∗, 2  i=1 ∇(kn+1 i−kn i)2L2(Ωi)≤c(δ2 1+1)δ2 2 ν3M 2  i=1 ∇(kn i−kn−1 i)2L2(Ωi).(3.6) P oo . The p oo o his heo em is made in se e al s eps. Fi s s ep. Choice o he es unc ion We fi s choose a pa icula es unc ion ϕiin he equa ions (2.7)–(2.9). Fo ha pu pose, we need o in oduce he special space H 1 2 00(Γ) (see [20], Chap. 1, Thm. 11.7 o ins ance). We also need o in oduce he ollowing ope a o . Le Ribe a con inuous ha monic li ing ope a o om H 1 2 00(Γ) o H1(Ωi), defined as ollows. Fo any ηin H 1 2 00(Γ), Riηbelongs o H1(Ωi), and sa isfies ⎧ ⎨ ⎩ −ΔRiη=0inΩ i, Riη=ηon Γ,and Riη=0onΓ i. Mo eo e , one has ∀η∈H 1 2 00(Γ),RiηH1(Ωi)≤cRηH 1 2 00(Γ),(3.7) whe e cR>0 depends only on Ωi. Acco ding o Hypo hesis 3.1,∀n∈N∗,k n i∈W1,3(Ωi), hen i s ace on Γ belongs o W2 3,3(Γ). Thus, by Sobole ’s injec ions, i belongs o H1 2(Γ). Fu he mo e, kn i=0onΓ i, henkn i|Γbelongs o H 1 2 00(Γ). The idea consis s in choosing he es unc ion ϕiequal o (kn+1 i−kn i)−Ri(kn+1 i−kn i)inequa ion(2.9)a s eps nand n+ 1. Then, we make he diffe ence be ween bo h ob ained equa ions, and sum upon i=1,2. We find ν 2  i=1 ∇(kn+1 i−kn i)2 0,Ωi≤ 7  j=1 Ij,(3.8) 700 T. CHAC ´ ON REBOLLO ET AL. whe e I1= 2  i=1Ωi αi(kn i)(|∇un+1 i|2−|∇un i|2)(kn+1 i−kn i)dx , I2= 2  i=1Ωi (αi(kn i)−αi(kn−1 i))|∇un i|2(kn+1 i−kn i)dx , I3= 2  i=1Ωi (γi(kn i)−γi(kn−1 i))∇kn i·∇(kn+1 i−kn i)dx , I4= 2  i=1Ωi (γi(kn i)−γi(kn−1 i))∇kn i·∇Ri(kn+1 i−kn i)dx , I5= 2  i=1Ωi αi(kn i)(|∇un+1 i|2−|∇un i|2)Ri(kn+1 i−kn i)dx , I6= 2  i=1Ωi (αi(kn i)−αi(kn−1 i))|∇un i|2Ri(kn+1 i−kn i)dx ,and I7= 2  i=1Ωi γi(kn i)∇(kn+1 i−kn i)·∇Ri(kn+1 i−kn i)dx . Second s ep. Es ima es o Ij,1≤j≤7 Es ima ion o I1.We w i e |∇un+1 i|2−|∇un i|2=∇un+1 i−un i·∇un+1 i+un i, and use Hypo hesis 3.1,and ela ion (2.1). Thanks o he Sobole embedding o H1(Ωi)in oL6(Ωi)and omH¨olde and Poinca ´e-F ied ichs inequali ies, we ob ain Ωi αi(kn i)(|∇un+1 i|2−|∇un i|2)(kn+1 i−kn i)dx ≤δ1∇(un+1 i−un i)L2(Ωi)dun+1 iW1,3(Ωi)d+un iW1,3(Ωi)dkn+1 i−kn iL6(Ω) ≤Mcδ1∇(un+1 i−un i)L2(Ωi)d∇(kn+1 i−kn i)L2(Ωi), whe e cis a posi i e cons an , depending only on domains Ωi. To simpli y he calcula ions, we in oduce a posi i e numbe βwhich we shall fix la e . Acco ding o Young’s inequali y 1 βa2+βb2≥2ab, ∀a, b ∈R,and ∀β>0,(3.9) we ob ain Ωi αi(kn i)(|∇un+1 i|2−|∇un i|2)(kn+1 i−kn i)dx≤ν β∇(kn+1 i−kn i)2 L2(Ωi)+βM2δ2 1c2 ν∇(un+1 i−un i)2 L2(Ωi)d. AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL 701 Summing upon i=1,2, and due o he ela ion (3.3) om Lemma 3.3, he e exis s a posi i e cons an c1, depending only on Ωi,αiand M, such ha I1≤c1βδ2 1δ2 2M2 ν3 2  i=1 ∇(kn i−kn−1 i)2 L2(Ωi)+ν β 2  i=1 ∇(kn+1 i−kn i)2 L2(Ωi).(3.10) Es ima ion o I2and I3.Using he same a gumen s we used o es ima ion o I1, he e exis s wo posi i e cons an s, depending only on Ωi,γ iand M, such ha I2≤c2βδ2 2M2 ν 2  i=1 ∇(kn i−kn−1 i)2 L2(Ωi)+ν β 2  i=1 ∇(kn+1 i−kn i)2 L2(Ωi),(3.11) and I3≤c3βδ2 2M2 ν 2  i=1 ∇(kn i−kn−1 i)2 L2(Ωi)+ν β 2  i=1 ∇(kn+1 i−kn i)2 L2(Ωi).(3.12) Es ima ion o I4.We ecall ha I4= 2  i=1Γi (γi(kn i)−γi(kn−1 i))∇kn i·∇Ri(kn+1 i−kn i)dx . Le us apply he Mean Value Theo em o he unc ion γi,use ela ion(2.2)andH¨olde inequali y. We find Γi (γi(kn i)−γi(kn−1 i))∇kn i·∇Ri(kn+1 i−kn i)dx≤δ2kn i−kn−1 iL6(Ωi)∇kn iL3(Ωi)∇Ri(kn+1 i−kn i)L2(Ωi). The con inui y o he li ing ope a o (3.7)andHypo hesis3.1 imply Γi (γi(kn i)−γi(kn−1 i))∇kn i·∇Ri(kn+1 i−kn i)dx≤cRδ2Mkn i−kn−1 iL6(Ωi)kn+1 i−kn iH 1 2 00(Γ). Acco ding o he con inui y o he canonical injec ion om H1(Ωi) oL6(Ωi), he con inui y o he ace ope a o om Xi o H 1 2 00(Γ), and using Young’s inequali y (3.9), he e exis s a posi i e cons an c4>0, depending only on Ωi,γ i,and M, such ha Γi (γi(kn i)−γi(kn−1 i))∇kn i·∇Ri(kn+1 i−kn i)dx≤c4βδ2 2M2 ν∇(kn i−kn−1 i)2 L2(Ωi)+ν β∇(kn+1 i−kn i)2 L2(Ωi). Summing on i=1,2, we find I4≤c4βδ2 2M2 ν 2  i=1 ∇(kn i−kn−1 i)2 L2(Ωi)+ν β 2  i=1 ∇(kn+1 i−kn i)2 L2(Ωi).(3.13) Es ima ion o I5.We ha e I5= 2  i=1Ωi αi(kn i)(|∇un+1 i|2−|∇un i|2)Ri(kn+1 i−kn i)dx . 708 T. CHAC ´ ON REBOLLO ET AL. embedding om W1, 0(Ωi) oH1 0(Ω), we can w i e o m≥m0 ϕm i−ϕiH1 0(Ωi)≤η 3·(4.3) Thanks o he iangula inequali y, we ob ain Ωi γi(kn i)∇kn+1 i·∇ϕidx−Ωi γi(ki)∇ki·∇ϕidx≤Ωi γi(kn i)∇kn+1 i·∇ϕidx−Ωi γi(kn i)∇kn+1 i·∇ϕm idx +Ωi γi(kn i)∇kn+1 i·∇ϕm idx−Ωi γi(ki)∇(ki)·∇ϕm idx +Ωi γi(ki)∇(ki)·∇ϕm idx−Ωi γi(ki)∇(ki)·∇ϕidx . (4.4) H¨olde inequali y implies Ωi γi(kn i)∇kn+1 i·∇ϕidx−Ωi γi(ki)∇ki·∇ϕidx≤δ1∇kn+1 iL2(Ωi)∇(ϕm i−ϕi)L2(Ωi) +Ωi γi(kn i)∇kn+1 i·∇ϕm idx−Ωi γi(ki)∇(ki)·∇ϕm idx +δ1∇ki0∇(ϕm i−ϕi)L2(Ωi). Using ela ions (4.2)–(4.3), and he s ong con e gence o he sequence (kn i)nin H1(Ωi), we deduce ha o any unc ion ϕi∈W1, 0(Ωi), we ha e lim n→∞ Ωi γi(kn i)∇(kn+1 i)·∇ϕidx=Ωi γi(ki)∇(ki)·∇ϕidx. To show he second equa ion o he ela ion (4.1), we w i e Ωi αi(kn i)|∇un+1 i|2−αi(ki)|∇ui|2ϕidx ≤Ωi αi(kn i)|∇un+1 i|2−|∇ui|2|ϕi|dx+Ωi|αi(kn i)−αi(ki)||∇ui|2|ϕi|dx, hanks o Hypo hesis 3.1, he Sobole embedding om H1(Ωi) oL6(Ωi)andH¨olde inequali y, we can w i e o all ϕi∈D(Ωi), Ωi αi(kn i)|∇un+1 i|2−αi(ki)|∇ui|2ϕidx ≤δ1ϕi∞∇un+1 iL2(Ωi)d∇uiL2(Ωi)d∇(un+1 i−ui)L2(Ωi)d+cM2δ2ϕi∞kn i−kiH1(Ωi), whe e cis a posi i e cons an , depending only on he Ωi. Due o he s ong con e gence in H1o he sequences (un i)nand (kn i)n o uiand ki, Ωiαi(kn i)|∇un+1 i|2−αi(ki)|∇ui|2ϕidx≤η 3· AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL 709 Le now ϕi∈W1, 0(Ωi). The densi y o D(Ωi)inW1, 0(Ωi), implies ha he e exis s a sequence (ϕm i)min D(Ωi), such ha ϕm i−ϕiL3(Ωi)≤c(Ωi)ϕm i−ϕiW1, (Ωi)≤η 3, whe e c(Ωi) only depends on Ωi. Finally, using he iangula inequali y, we ob ain Ωiαi(kn i)|∇un+1 i|2−αi(ki)|∇ui|2ϕidx≤Ωiαi(kn i)|∇un+1 i|2−αi(ki)|∇ui|2ϕm idx +Ωi αi(kn i)|∇un+1 i|2(ϕm i−ϕi)dx +Ωi αi(ki)|∇ui|2(ϕm i−ϕi)dx . Thus, Ωiαi(kn i)|∇un+1 i|2−αi(ki)|∇ui|2ϕidx≤αi∞∇un+1 i2 L3(Ωi)dϕm i−ϕiL3(Ωi) +ε 3+αi∞∇ui2 L3(Ωi)dϕm i−ϕiL3(Ωi). We deduce ha ∀ϕi∈W1, 0(Ωi),lim n→∞ Ωi αi(kn i)|∇un+1 i|2ϕidx=Ωi αi(ki)|∇ui|2ϕidx. Thi d s ep. Bounda y condi ions o TKE on Γ In his s ep, we show ha ki=λ|u1−u2|2on Γ. Conside he ollowing iangula inequali y λ|u1−u2|2−kiH1 2(Γ) ≤λ|un 1−un 2|2−kiH1 2(Γ) +λ|un 1−un 2|2−λ|u1−u2|2H1 2(Γ). Due o he s ong con e gence o (kn i)n o kiin H1(Ωi), and hanks o he con inui y o he ace ope a o om H1(Ωi) oH1 2(Γ), he sequence kn icon e ges s ongly o kiin H1 2(Γ). Fu he mo e, kn i=λ|un+1 1−un+1 2|2 on Γ. Thus, lim n→∞ λ|un+1 1−un+1 2|2=ki, s ongly in H1 2(Γ). Le us now p o e ha lim n→∞ λ|un 1−un 2|2−λ|u1−u2|2H1 2(Γ) =0. Using he iden i y a2−b2=(a−b)(a+b) and Lemma 3.5,weob ain λ|un 1−un 2|2−λ|u1−u2|2H1 2(Γ) ≤λ||(un 1−u1)−(un 2−u2)||H1 2(Γ) || (un 1+u1)−(un 2+u2)||W1−1 3+ε,3+ε(Γ)d, so, λ|un 1−un 2|2−λ|u1−u2|2H1 2(Γ) ≤λun 1−u1H1 2(Γ) +un 2−u2H1 2(Γ)un 1+u1W1−1 3+ε,3+ε(Γ)d+un 2+u2W1−1 3+ε,3+ε(Γ)d. Due o he con inui y o he ace ope a o s om W1,3+ε(Ω1)d o W1−1 3+ε,3+ε(Γ)dand om H1(Ωi) oH1 2(Γ), and by Hypo hesis 3.1, he e exis s a posi i e cons an c, ha only depends on he domains Ωi, such ha λ|un 1−un 2|2−λ|u1−u2|2H1 2(Γ) ≤cMλ un 1−u1H1(Ω1)+un 2−u2H1(Ω2). 710 T. CHAC ´ ON REBOLLO ET AL. Finally, due o he s ong con e gence o (un i)n o uiin H1(Ωi), we deduce ha lim n→∞ λ|un 1−un 2|2−λ|u1−u2|2H1 2(Γ) =0. Consequen ly, ki=λ|u1−u2|2on Γ. This, finishes he p oo o his heo em.  5. Nume ical expe imen s To conclude his pape , we use he algo i hm in oduced o sol e he in e ac ion o he ocean and he a mosphe e in a simplified geome y. The disc e iza ion is pe o med using a Spec al me hod based on Legend e polynomials (see [2,10]o [11] o ins ance) ha we ha e implemen ed in F eeFEM3D4. The algo i hm p esen ed in his pape is nonlinea , hus i canno be used “as is”. Va ious s a egies could be employed o ea his nonlinea i y (New on o fixed poin algo i hms, e c.). Acco ding o he mono onic na u e o he nonlinea ic ion bounda y condi ion on he bounda y Γ, we choose o linea ize he e m Γ|un+1 i−un+1 j|(un+1 i−un+1 j)· idτ o P oblem 1. We eplace i by Γ|un i−un j|(un+1 i−un+1 j)· idτ. This e m is linea and s ill mono onic in he unknowns (un+1 1,un+1 2), due o p ope y (3.4). The p oblems o (un+1 1,p n+1 1)and(un+1 2,p n+1 2) s ill a e coupled bu he o e all p oblem is linea and admi s a unique solu ion: Ob ain (un+1 i,p n+1 i,kn+1 i)∈Xi×L2(Ωi)×W1, (Ωi), such ha ∀( i,q i,ϕ i)∈Xi×L2(Ωi)×W1, 0(Ωi), ai(kn i;un+1 i,∇ i)+bi( i,p n+1 i)+κiΓ|un i−un j|(un+1 i−un+1 j)· idτ=Ωi i· idτ, (5.1) and bi(un+1 i,q i)=0.(5.2) To pe o m a somewha ealis ic compu a ion, we conside u bulen iscosi ies αiand γiwi h he s uc u e ν +√k. We conside he da a epo ed in [5]: •Geome y: –Ω1=]0,5[ ×]0,1[ ×]0,1[ desc ibes he a mosphe e; –Ω1=]0,5[ ×]0,−1[ ×]0,1[ is he ocean. •Physical da a ( aken om [5]): –γ1(k1)=3×10−3+0.277 ×10−4√k1; –γ2(k2)=3×10−2+0.185 ×10−5√k2; –αi(·)=γi(·), o i=1,2. •F ic ion coefficien s (coming also om [5]): –κi=10 −3, o i=1,2; and –λ=5×10−2. These da a co espond o an ai -sea flow, each modeled by a simplified TKE-mixing laye u bulence model. The mixing leng hs a e calcula ed by wall laws. The ic ion coefficien s, in hei u n, a e jus en a i e. The physical uni s a e in MKS sys em. The eloci y bounda y condi ions imposes ha u1=0on Γ1 ˜ Γ1, u1=(1,0,0) on ˜ Γ1and u2=0on Γ2,whe e˜ Γ1is he uppe ace (y=1)o Ω 1. These se ings a e chosen in o de o c ea e a d i en ca i y-like flow in Ω1. One expec s o gene a e ano he d i en ca i y-like flow in Ω2 o a ing in he opposi e sense. 4h p://www. ee em.o g/ 3d/. AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL 711 (a) u1on he plane z=0.5(b)u1on he plane x=2.5 (c) u2on he plane z=0.5(d)u2on he plane x=2.5 Figu e 1. Veloci y fields on cu ing planes. (a) k1on he plane z=0.5(b)k2on he plane z=0.5 Figu e 2. Tu bulen Kine ic Ene gy on cu ing plane. The esul s a e ob ained using a (PN)3×PN−2disc e iza ion o (ui,p i) o a oid spu ious modes o he S okes p oblem. The TKE is disc e ized using a PNspace. Fo his pa icula simula ion we ha e chosen he ollowing deg ees o uiand ki: 28 in di ec ion xand 8 in di ec ions yand z. Compu ed eloci y fields uiand he TKE kia e ep esen ed in Figu es 1and 2. The esul s a e quan i a i ely co ec : The a mosphe e flow gene a es a d i en-lid like flow in he ocean, due o he bounda y condi ions on he eloci y a z= 0. Also, he e is gene a ion o TKE a z= 0, again due o he TKE gene a ion bounda y condi ions. 712 T. CHAC ´ ON REBOLLO ET AL. 1e-14 1e-12 1e-10 1e-08 1e-06 0.0001 0.01 1 0 1 2 3 4 5 6 7 8 9 u1 u2 k1 k2 Figu e 3. Con e gence his o y: compu ed L2no m o he diffe ence o successi e i e a es. Figu e 3shows he expec ed exponen ial con e gence a e o he algo i hm due o i s con ac i eness. No e ha his does no depend on he ype o disc e iza ion (see [24] whe e Fini e Elemen app oxima ions ha e also been used). 6. Conclusion In his pape , we ha e p esen ed and analyzed a nume ical scheme o he app oxima ion o a model o wo s eady u bulen fluids wi h coupling a he in e ace. This is a simplified model o he a mosphe e-ocean in e ac ion, whe e we ha e neglec ed Co iolis o ces and buoyancy effec s, bu ha e kep se e al non-linea in e ac ions ac oss he common bounda y. The p oposed scheme is mainly linea , a mono one nonlinea i y being jus kep a he in e ace be ween he fluids. We showed he con e gence o he iple (un i,p n i,kn i), o easonable hypo hesis on he egula i y o he eloci y and he u bulen kine ic ene gy. This con ibu ion ends wi h some nume ical esul s, in good ag eemen wi h he heo e ical expec a ions. No ably, he exponen ial con e gence ha appea ed h ough he con ac i eness o he sequences (un i)nand (kn i)nis ound in ou es s. Se e al ex ensions o his wo k can be conside ed. To name a ew, aking in o conside a ion aniso opic diffusion and Co iolis o ces (mo e ealis ic) a e s aigh o wa d gene aliza ions o he p esen analysis. Also, swi ching o uns eady incomp essible flows should be possible using ou app oach. Taking in o accoun buoyancy effec s is mo e echnically in ol ed, bu i should also be possible, simila ly o he ex ension o he s anda d analysis o incomp essible Na ie –S okes o buoyancy effec s. Acknowledgemen s. The au ho s a e g a e ul o Ch is ine Be na di o he help and encou agemen s. Re e ences [1] J.J.F. Adams and R.A. Fou nie , Sobole spaces. Second edi ion, Pu e and Applied Ma hema ics Se ies, Else ie /Academic P ess (2003). [2] C. Be na di and Y. Maday, App oxima ions spec ales de p obl`emes aux limi es ellip iques,Ma hema ics & Applica ions 10. Sp inge -Ve lag (1992). [3] C. Be na di, T. Chac´on Rebollo, R. Lewandowski and F. Mu a , A model o wo coupled u bulen luids. I. Analysis o he sys em, in Nonlinea pa ial diffe en ial equa ions and hei applica ions, Coll`ege de F ance Semina ,Vol.XIV (Pa is, 1997/1998), S ud. Ma h. Appl. 31, Ams e dam, No h-Holland (2002) 69–102. AN ITERATIVE PROCEDURE TO SOLVE A COUPLED TWO-FLUIDS TURBULENCE MODEL 713 [4] C. Be na di, T. Chac´on Rebollo, R. Lewandowski and F. Mu a , A model o wo coupled u bulen luids. II. Nume ical analysis o a spec al disc e iza ion. SIAM J. Nume . Anal. 40 (2003) 2368–2394. [5] C. Be na di, T. Chac´onRebollo,M.G´omez M´a mol, R. Lewandowski and F. Mu a , A model o wo coupled u bulen luids. III. Nume ical app oxima ion by ini e elemen s. Nume . Ma h. 98 (2004) 33–66. [6] C. Be na di, T. Chac´on Rebollo, F. Hech and R. Lewandowski, Au oma ic inse ion o a u bulence model in he ini e elemen disc e iza ion o he Na ie -S okes Equa ions. Ma h. Mod. Me h. Appl. Sci. 19 (2009) 1139–1183. [7] H. B ezis, Analyse Fonc ionnelle : Th´eo ie e Applica ions. Collec ion “Ma h´ema iques Appliqu´ees pou la Maˆı ise”, Masson (1983). [8] F. B ossie and R. Lewandowski, Impac o he a ia ions o he mixing leng h in a i s o de u bulen closu e sys em. ESAIM: M2AN 36 (2002) 345–372. [9] K. B yan, A nume ical me hod o he s udy o he ci cula ion o he wo ld ocean. J. Compu . Phys. 4(1969) 347–369. [10] C. Canu o, M.Y. Hussaini, A. Qua e oni and T.A. Zang, Spec al me hods – Fundamen als in single domains. Sp inge , Be lin, Ge many (2006). [11] C. Canu o, M.Y. Hussaini, A. Qua e oni and T.A. Zang, Spec al me hods – E olu ion o complex geome ies and applica ions o fluid dynamics. Sp inge , Be lin, Ge many (2007). [12] S. Del Pino and O. Pi onneau, A ic i ious domain based on gene al pde’s sol e s, in P oc. ECCOMAS 2001, Swansea, K. Mo gan Ed., Wiley (2002). [13] V. Gi aul and P.-A. Ra ia , Fini e Elemen Me hods o Na ie -S okes Equa ions, Theo y and Algo i hms. Sp inge -Ve lag, Ge many (1986). [14] P. G is a d, Ellip ic P oblems in Nonsmoo h Domains,Monog aphs and S udies in Ma hema ics 24. Pi man (Ad anced Publishing P og am), Bos on, USA (1985). [15] E. Hebey, Nonlinea analysis on mani olds: Sobole spaces and inequali ies,Cou an Lec u e No es 5. Ame ican Ma hema ical Socie y, USA (1999). [16] B.E. Launde and D.B. Spalding, Ma hema ical Modeling o Tu bulence. Academic P ess, London, UK (1972). [17] J. Lede e and R. Lewandowski, A RANS 3D model wi h unbounded eddy iscosi ies. Ann. Ins . H. Poinca ´eAnal.Non Lin´eai e 24 (2007) 413–441. [18] R. Lewandowski, Analyse Ma h´ema ique e Oc´eanog aphie. Collec ion Reche ches en Ma h´ema iques Appliqu´ees, Masson (1997). [19] R. Lewandowski, The ma hema ical analysis o he coupling o a u bulen kine ic ene gy equa ion o he Na ie -S okes equa ion wi h an eddy iscosi y. Nonlinea Anal. 28 (1997) 393–417. [20] J.-L. Lions and E. Magenes, P obl`emes aux limi es non homog`enes e applica ions 3,T a aux e Reche ches Ma h´ema iques 20. Dunod, Pa is, F ance (1970). [21] B. Mohammadi and O. Pi onneau, Analysis o he k-epsilon u bulence model. RAM: Resea ch in Applied Ma hema ics. Masson, Pa is (1994). [22] J. Pique , Tu bulen Flows, Models and Physics. Sp inge , Ge many (1999). [23] D.C. Wilcox, Tu bulence Modeling o CFD. Six h edi ion, DCW Indus ies, inc. Cali o nia, USA (2006). [24] D. Yakoubi, Analyse e mise en œu e de nou eaux algo i hmes en m´e hodes spec ales.Ph.D.Thesis,Uni e si ´e Pie e e Ma ie Cu ie, Pa is, F ance (2007).