Inn of Sa! pe 171. N 2.198 A BGK Model for Small Prandt! Number the Navier-Stokes Approximation rages Boucut! and Belt Pertare? frame nodund mip te cane opr by «noi [Elon fe toon int ean oe Pn mi 4, IwTRooucTION ‘The mot accepted model which describe the evolution af the density a ‘arefed pie the Boltmann equation sysevrebourn w shee f(t 9) the dtebuton function, 1 the vine viable the space ‘arable, and the elosty variable. Because of the complet ofthe Botemens ellsion term QU fl, any authors tee to subtiute simpler opeatr, as inthe BGK mode! introduced by Bhatogar. Gross {nd Kook or Welander™ where the operator Qi replaced by Wawra e PRIMUS, du Rah Sag 4071 Orns een Frc ‘lemon ts nto ft BF, Oe an 2Unaly + taken proportional othe macroscopic density pl, x) and {MLA} isthe Maxwelianasociated to (ae btow for presse defiitons. This model equation presrves the base propenes of the Boltzmany ‘equation: conservation of us, momentum, and energy, 2nd iasiption of entropy alo ives, inthe Mid limit the Euler equations. ‘hs paper we ae treed in the Naver Stoke equations derived ‘kom (2) trough a Chapman-Easkog expansion ™ Many author hive worked on this subject for the Boltzmann equation (ee for tastance {els 1 3,5,89, ad Te) and Wis wellknown that the resulting Pranl umber’ ior monatomic gas)'is Pr=2/3. For the BGK mode, Cercgnani pointed out that the ascisted Prandtl number is | and consequently ase the problem of fading a model wbich woul piv other rand numbers while keeping the Hseorem. “There are oter motivations for this question. Fist, i ot easy to find in the Boltzmann equation (1) the ess section adapted to 8 ves, rand number. Second, such model could have numeral applications, asin the work of Puls and Reitz sine the BGK model simpler, 'o approximate than the Bolzmana caution ‘We now preset sich « model We inoduce @ multiplier (6) while =} and we et as eoV,Svy(s—M)=0 o where Non Meee {5 the Masvelin of desty 2}, mean velosky a} and temperate Ta) Thee macrscpi oui a chown flo Pl we doe Pan at Top Onrap edu + NT =[ MheWPifWxed — sat , sant sss eren (Bete) ‘Then 9, and Tare chosen so tht LeMetinttran[ anterid oO {80K Mola Navin StkaeAperximation te Ta fact it wil urn ot that (py Hy T/) and (pots T) are 1 close and thos, for the NavieeSlokes approtimation, thee two macroscopic ‘arable ae equivalent Final, we wll choose Hex) 0/67) ® and $e rltd to the behavior in T of the viscosity and heat conduction terms of Naver Stokes equations “The Prada number Pris governed by the choie of 4() Here we 3S OKC SAW EC < 40 ° Ion) = Ae) 0) Me itd ns 1 Jaen on Then _2Ms2) 22). 92 ay We have uted the notation Malu) wo on A cifernce rom the model proposed in re. 4 that, a the me ie rath epoes 100, feonerges to a Maxwell, thus recovering the casi exiy lit “The fs of the paper is organized as follows, In Seston 2 we derive ‘he relation (12) forthe Prana mumbern the Navte-Stokesappcaime ton. Then we prove in Saion 3 that we cam inde compute (p.m) i 2 unique way rom the Felation (7) We also introduce 4 "varitional™ eine whch gives the H-heorem. Seton & i devoted to an existence {nd stability proof for Eq (3) when ¥=1, wing the avenging lemma Finally, we show in Setion 5 tht the whole terval}, 1] forthe Prandtl umber can be reached by appropriate choices of ly2, DERIVATION OF THE NAVIER-STOKES EQUATIONS In thie section, we show the reltonip between the Kinetic model v4 the hydrodynamic equation. We dope a ervtion bic dlrs Slightly from the casical Chapman-Enskog expansion'™*"" We have {ound this presottion in Deshpande"! and st scems simpler for out purpose. "This ection ie organized as follows, We fret recall Bow 10 recover the "Navier-Stokes equations. ‘Then we prove that” (9,m 7) = (iyeanT))+ Oe) Tht allows ur to compute the Navier-Soker ‘cuete deduced fom our BGK. model. Thea we conte this section with some emai 2.1. Tho Navier-Stokes Equations We considera rtion fof the equation aS +0V.S =o me as) ith he eink (4) Sine tii [tonwrreano the map ues py Ts ton ap/+diviga=0 US) Alomntdinieny@u+Pl=0 (16) sole) oo (ME 8B wero on “he mix Pande vector @ ate dined by Pee [leu lenap fae on anifienyP emu sae sy Finally, w@ denotes the symmeti mateix wih coeicients wy, and iva) the vector of components 24a, “The system (15)-(17) the compioeble Nvir-Stkes stem fora Peoptin- tty (eZee) e o=-x0p9F, an he ee ety at oe 24 Bet a ‘The Prandtl number isthe ratio a 122. A Proliminary Result In order to compute P and Q up o« ate goes 100) when fsatises the BGK equation (14), we jst notice that Jo M =o.) on and thus, esing (14 gai, IoMn~ eS e VS) oo Fyre Vaan soe) 25) We will show tat eplasing fin (18) and (19) bythe expression given in tems of trough (25) yes the appropriate form of P and. ‘We need a preliminary rsa: Lemma 1. We have [en nry-aneenore) a In other words, (nT) dls fom (p,m T) by term of onder ay