Twenty-Third Symposium on Naval Hydrodynamics (2001) / Chapter Skim
Currently Skimming:
Large-Eddy Simulations of Turbulent Wake Flows
Large-Eddy Simulations of Turbulent Wake Flows
Pages 579-598
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 579... ...
smell eddies me modeled hus v ry flme grids hav to be applied m order to lesolv She bo mdary layer near the wall where the turbulence length scales tend to Pro es the the wall is aproahed However, m some cpplicctions, like bubble dynamics modeling, it is still necessary to ~esolv coherent -I w shuctmes - large turbulent vortices Ed eddies in RANS simulations, e peciclly those using two-equation models such es k—~ model, these m teddy -I w fectmes me usually smeared out he prediction of bubble population m She wake is importat m controlling She signature of surface ships Since it c m be computatiorurlly proh~bitiv to mchde the ship hull ad the wake in LES, it would be desirctle for She purposes of pure wake simulations to start the computations somewhere m the near wake exchding the shipbody his technique, called by of her researchers the Initial Deb Phne ,Ir)
|
From page 580... ...
Comparisonbetweff~ LES md experiments or tANS hav been mcde Con cl 3sions md r commendmg for f 3tme work are giv n et the end 2 MATHEMATICAL / NUMERICAL MODELS 2.1 Navier-Stokes Solver tion of the gov rnmg equations md 6he SGS models which are 3sed m 6his t 3dy The LES code we 3se was origir~lly developedby Z mg et cl (1994) The spaticlly flltered fl w conservation equations are I aui =o a~j Ou~ _ a at a~ ptj = q~qj _ p6~j T,j = U,uj — TTj — 3 T t = —2VTSI; o (1)
|
From page 581... ...
scheme is cpplied to discr tize 6he cor ctive terms C nbal differ n ing scheme is used for 6he odher terms With QUiCK scheme, no subg id-sccle (SGS) model is used while for centrcl dffferen mg scheme Smogarinskymodel (Smcgormsky,1963)
|
From page 582... ...
, paticl md t mporcl varictions of u~ follow G mssi m distribution wi6h charateri tic length md time-sccles of / z, how v r, other distributions m be used to simulate dffferent tmbulence spechc More details of this method cm be fo md m Smi novet cl (2000b) 3 APPLICATIONS he pseudo~mdom flow field genercted by the RFG medhod is cdded to 6he mea flow to established 6he bo mdary condition et the inlet pkme he ov cll procedme is cpplied to two sigmificmtly dffferent cases, rumely flct plate wake md c ship model wake Since w try to zvoid solving for the fl w aro md 6he bodies, w tart the computatiorurl domcin from 6he i flow bo mdary(initial date plane)
|
From page 583... ...
The ~esolv d turbulence Emetic energy Ti E) for different schemes mddifferentg id sins is show mFig 12 The tANS solution is 31 so show in This picture for comparison The ~esolv d TKE from flme g id is higher th m that of the coarse g id as expected C ntrcl dffferencmg discretization with Smagormsky model giv s better resuits th m the ocher t hemes From this flame it also c m be seen that there is not sig dic mt dffference in fihe resuits when the QUiCK sch me is used with or wifihout SGS model hforeov r, fihe resolv d TKE is I wer th m that of central dffferencmg wifih Smagorinsky model It me ms Shut QUICK scheme giv s ev n higher mmmerical diffusion fi m fihe Smagormsky model This is mostly due to g id resolution For mme detailed information about fihe comparisons of m merical schemes md subg id-scale m odels the reader is refened Shi et al (2000a)
|
From page 584... ...
A conesponding m cc m the prticity contop plot is clso ippluded m Fig 22 for ~eferep he cross s ption is et L= 1 2 Havingno interpolation, it seems6mt inthe velocity vector plot the Weak vortices cmm not be seen dp to pcle dfffereppe, wherecs the mtemolated contop s plots do show smell shuptpses Figmes 21 mmd 22 show cgam thmt mpch of 6he vo ticity is coppentmted p m the free sp fcce mmd the~e m two Imge counter rot dmg vortices on two sides of the wake Ac tmdioc oi the hip wake how t also w ilable d htp://cid.mae.wm.ed /ddpwshe 5 CONCLUSIONS AND FUTURE WORK A combip d LES scheme mmd mm tFG cpprocch spitctle for wake fiows is briefiy de pribed he Wdke fiow of c fict pkte is psed es c mlidation case Both the memm fipw mmd tp bpleppe intensities m compm d with 6he experimenwl ~esplts A good cg cement hms been obtamed R gmding ship wake simpktions it is diffcult to d cw mmy defimite copplpsions et 6his pomt in the cbsff~p of experimental daw How ver, 6he ~espits me very ~ecsopmble mmd it memms op cpprocch is c victle dpprocch m 6 mt it is capable of ccptm ing the mo d ep rgetic pmstecdy, tp bplent vo tices mmd or cddies p~esent in the Wdke his st dy shows 6h d with recsopmble k ge g id nodes LES of ship waker hmr very good pro pects Never6heless, f 11 sccle LES cclcpktions of ship fiow me still imposs~ble b pmmse of the limitation of computer resop p s mmd may requi~e mmother 10 or 20 yem s Lm sson et cl,1998) Appording to Hymmm Hymmm, 1998)
|
From page 585... ...
Smirnov, A, md C lik,1 2000c, Large eddy simulations of particle-kden tmbulent wakes using c r mdom fl w genemtion techmique, m ONR 2000 Fnee Surfoce Turbulence and Bubbly Flow Wwk shop, pp 13 1 13 7, Cclifornic lostitute of Techmology, Pcsaderur, CA Smcgormsky,J 1963, Generclci culationexperiments with the prim itive equctions, part i the basic experiment, Monthy Weo6herRex 91, 99 Smi nov, A, Shi, S md C lik, 1 2000c, R ndom Flow G menotion Technigue in la ge Eddy Simula tions and Ponticle Dyn mics Modeling, submitted to Jourm~l of Fluid E gineermg Smi nov, A, Shi, S
|
From page 586... ...
md lioiett, J.: 1994, A nonstaggered g id, fractional step method for timedependent income Ale r~vier-stokes equations m curvilinear coordinates, Journal of Comp to do. al Physics 114,18
|
From page 587... ...
~ ~ 1 1 1 1 1 265000 270000 265000 270000 iterations iterations Figure 2: Temporal history of the instantaneous velocity components at different points on the center line of the flat plate wake (x = 0 is the inflow boundary)
|
From page 588... ...
0 0.1 0.2 0.3 0.4 0.5 0.6 x (m) Figure 5: Kinetic energy profile along the center line in the wake of flat plate I ~/ /00 >' 200 175 150 125 50 - Experiment (Ramaprian et al., 1981 )
|
From page 589... ...
A photograph showing the large coherent structures in the flat plate wake (Wygnanski et al., 1986) Figure 7: Comparison of the turbulence structures between LES and experiments
|
From page 590... ...
on a x-z plane parallel to * ee surface at y/L = - 0.01 in the ship wake
|
From page 591... ...
u 0.945' 0.927' 0.908' 0.890' 0.871 ~ 0.853' 0.834( 0.816' 0.797 0.779 0.760 0.742~ 0.723E 0.705 0.686E 0.668( 0.649' 0.631 ( 0.612' Figure 1O: The streamwise velocity contours on a x-z plane parallel to free surface at y/L =—O.01 in the ship wake ' ~- 0.2 n ~ O 0.4 ~.
|
From page 592... ...
I~lf 1\ t~ . ~ 1 \ ~ Figure 13: The velocity vectors on y-z plane at x/L = 0.2 in the ship wake -0.05 -0.06 _ -n nR ~ ~ , ~ , ~ , -0.05 0 0.05 Figure 14: The velocity vectors on y-z plane at x/L 0.6 in the ship wake -0 02 -n nn Figure 15: The velocity vectors of finer grid solution on y-z plane at x/L = 0.2 in the ship wake
|
From page 593... ...
-... -0.07 _ -0.08 -o 05 0 0.05 Figure 18: The velocity vectors on y-z plane at x/L = 1.4 in the ship wake
|
From page 594... ...
on y-z plane at x/L = 0.2 in the ship wake (finer grids are applied on both y and z directions) 0.025 0.05 0.075 ~~ 10.66a 9.6499 8.6396 7.6293 6.61 gal 5.6087 4.8005 3.7902 2.7799 1.9043 1.0287 0.4225 -0.01 56 -0.4625 -1 .1939 -1 .9348 -2.7430 -3.6859 -4.561 5 -5.4371 -6.31 27 -7.3230 -8.3333 -9.3436 -1 0.353 ~ ~ PR4 Figure 21: The vorticity contours (fox)
|
From page 595... ...
could be rented to grid resolution AUTHOR'S REPLY We appreciate Dr Hym m's summit y Ed comments on this paper Whet follows is c brief response to his questions QUESTIONS Question 1: "I would I ke to ask the mfhors how they would propose to characteri e this flow by c single lengfh-sccle, what f is length-sccle should be Ed how this lengfh-sccle should relate to grid resolution for LES calcubtiom " Though c con t mt was used in this study, in our method it is not necessary to use c single lengthsccle In fact, m She cunent version the algorithm is Greedy impl merited with the lengfh sccle parameter es c f notion of space which c m be computed fiom k Ed ~ or k Ed ~ it adds practically no extra computttiorLtl burden to the procedure There c m be c slight loss of accuracy in satisfying the inst mtaneous contimmity equation in c flow field generated with She RFG procedure associated with using i homogeneous turbulent length-sccle How ver, for She purposes of inlet plane initialization c smell violation of contimmity c m be tolerated since the flow solver will mtomaticclly adjust the velocities in She next plane such that the flow field sati fies continuity The lengfh scale used m the RFG procedure is cunently not related to grid resolution This might be c good idea to pursue m the future Question 2: I would like to ask ff the mthors' experience with LES cclcubtions using the QUICK scheme land lack of SGS model) could be rented to grid resolution A swer: From Fig 12 m the paper, it c m be seen Nat there is no signffi mt difference in She results when QU CIC scheme is used with or without SGS model Moreover, She resolved turbulent kinetic energy is lower th m that of central differencing with Smcgorinsky mod I This me ms that QU CIC scheme gives even higher diff sion thm Smcgormsky mod I Theoretically, the mmmericcl diffusion increases with increasing grid si e, so She low r resolution of sccle in the LES cclcubtions using the QUICK scheme (Ed k k of SGS model)
|
From page 596... ...
On p 4, col 2, the mthors comment upon He waviness of the velocity profile m Fig 3 They saythat this isbec mse theyhave averaged only clo g the centerline Instead of in the whole plane This is probably correct, but the same effect would of course be achieved if they increased She time Integration The reco h s show in Figs 3 -6 have been obtained with the QUiCK scheme md no SGS model The mthors say that they also have carried out simulations with central dffferencmg md She Smcgormsky model Whet about These results? It ..-m~ld he interesting to see c comparison A lot of contour plots are presented I don't thi k this is very mterestmg I would prefer to see m ore qu mtitative results Furthermore, I'm surprised that in She wake flow behind She ship model no comparison is mad wish experiments Why?
|
From page 597... ...
outc me of this comparison may be, LES technique provides extna i formation on unsteady krge-sccle component of the flowfield, Ed thereby provides c more consistent Ed necessary approach for mod hng Lcgr mgi m bubble dynamics m the wake which is the ultimate goal of the present work his, in our opinion, gives LES c definite cdv mtage over RANS methods in this particular application mea Answer to Question 3: Fig I included h re shows the treamwise velocity et the center line calculated by using QU OK scheme without subgrid sale model Ed central dffferencmg with Smcgorinsky mode As expect, the central differencing scheme gives better results compared to She QUICK scheme his is also tme for the turbulent intensities (not show ~ere I Answer to Question 4: Measurements of the flow-field m She wake of No y ship model 5415 is m on-going work Ed the experimental results are not available et this time he this respect our ship wake results represent c prediction rather than c post-diction Our results will be compared win h exper imental date es they bee ome cvailab le On the other hand our flct-plate computations presented in this paper w re specfficclly done to validate our model Ed provide justification for extending its application to th ship-wake it must be pointed out that the LES predictions w re done for c shear tree flat wall in teed of we y fiee surface he i tluence of waves on turbulence is usually not negligible
|
From page 598... ...
1. Modified comparison of predicted mean streamwise velocity at center line with measurements.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.