Lecture II: Granular Gases & Hydrodynamics Igor Aronson Materials Science Division Argonne National Laboratory Supported by the U.S. Department of Energy 1 Outline • • • • • • Definitions Continuum equations Transport coefficients: phenomenology Examples: cooling of granular gas Kinetic theory of granular gases Transport coefficients: kinetic theory 2 large conglomerates of discrete macroscopic particles Jaeger, Nagel, & Behringer, Rev. Mod.Phys. 1996 Kadanoff Rev. Mod. Phys.1999 de Gennes Rev. Mod. Phys.1999 non-gas inelastic collisions non-solid no tensile stresses Gran Mat non-liquid critical slope 3 Dropping a Ball • Granular eruption http://www.tn.utwente.nl/pof/ Loose sand with deep bed (it was fluffed before dropping the ball) 4 Group of Detlef Lohse, Univ. Twenta Granular Hydrodynamics • Let’s live in a perfect world -continuum coarse-grained description -ignore intrinsic discrete nature of granular liquid -ignore absence of scale separation • But, we include inelasticity of particles 5 Granular gases Definition: • Collections of interacting discrete solid particles. Under influence of gravity particles can be fluidized by sufficiently strong forcing: vibration, shear or electric field. • Granular gas is also called “rapid granular flow” 6 Comparison with Molecular Gases • Main difference – inelasticity of collisions and dissipation of energy • Common paradigm – granular gas is collection of smooth hard spheres with fixed normal restitution coefficient e v12 v12' (1 e )v12' kk v12 & v12' k are post /pre collisional relative velocities is the direction of line impact v1’ v2’ v1 v2 7 The Basic Macroscopic Fields • Velocity V • Mass density r (or number density n) • Granular temperature T (average fluctuation kinetic energy) Granular temperature is very different from thermodynamic temperature 8 Distribution Functions • Single-particle distribution function f(v,r,t) = number density of particles having velocity v at r,t • Relation to basic fields n(r , t ) dvf (v, r , t ) number density 1 V (r , t ) dvvf (v, r , t ) velocity field n( r , t ) 1 2 T (r , t ) dv ( v V ) f (v, r , t ) granular temperature n( r , t ) V, v, and r are vectors 9 Applicability of continuum hydrodynamics • Absence of scale separations between macroscopic and microscopic scales: Hydrodynamics is applicable for time/length scale S,L >> t,l t,l – mean free time/path l , T - granular temprature T For simple shear flow with shear rate g : Vx=gy Macroscopic time scale S=1/g Granular temperature T~g2l2 Restitution coefficient is in the t Vx prefactor t/S~tg~O(1) – formally no scale separation Restitution coefficient is a function of velocity Leo Kadanoff, RMP (1999): skeptic pint of view 10 Long-Range Correlations and Aging of Granular Gas • Inelasticity of collisions leads to long-range correlations • Example: fast particle chases slow particle elastic case – no correlation inelastic case (sticking) – correlations Usually particles don’t stick Lasting velocity correlations between different particles 11 Continuum equations • Continuity equation Vi Dn n 0 Dt ri where • Traditional form D V is the material derivative Dt t n nV 0 t Flux of particles J=nV, V – velocity vector Number of particles: Particles balance: N ndv ndv Ñ nVds nVdv t 12 Momentum Density Equations • force on small volume: ∫Fdv • acceleration: ∫nDV/dt dv • relation between force F and stress tensor sij: Fi s ij r j Compare in ideal fluid F p , p is pressure • Momentum balance: DVi s ij n - ng Dt rj where s ij is stress tensor, g -gravity 13 The Stress Tensor • Compare hydrodynamic stress tensor, Landau & Lifshitz Vi V j 2 Vl Vl s ij h ij x ij p ij rj ri 3 rl rl 144444444444444442 44444444444444443 Strain rate tensor • s Only appears c when contact ij duration > 0 Appears in dense flows, in granular gases ~0 h,x – first (shear) and second viscosities (blue term disappear in incompressible flow) • p – pressure (hydrostatic) • s ijc contact part 14 Note: energy is not conserved, but mass and momentum are Granular Temperature Equation • Detail derivation in L&L, Hydrodynamics Q j DT Vi n s ij G Dt rj rj energy sink (From inelastic collisions) heat flux shear heating • • • G ~n(1-e2) – energy sink term (absent in hydrodynamics) Q kT granular heat flux, k – thermal conductivity 15 Constitutive Relations: Phenomenology • relate h,k,G materials parameters (restitution e, grain size d and separation s) and variables in conservation laws n,V,T s d d • Typical time of momentum transfer t~s/u u ~T1/2 – typical (thermal) velocity • Collision rate = u/s 16 Equation of state • Pressure on the wall for s<<d using n~1/d3 mu 1 T p~ ~ const n 2 t d s • Volume V=N/n, N – total number of grains • s~V-V0; V0 – excluded volume p(V V0 ) const NT Analog of Van der Waal’s equation of state 17 Viscosity coefficient Vx(y) y • Two adjacent layers of grains x • shear stress from upper to lower layer s xy momentum transfer m DV u ~ d2 s collision rate u dVx ( y ) • velocity gradient DV/d ~dV/dy s xy const d r s dy u T 2 h~d r ~ • viscosity s n n0 2 r=m/d3 – density, n0- closed packed concentration 18 Thermal diffusivity • mean energy transfer between neigh layers The ratio of the two viscosities is constant, like in fluids • Mean energy flux muDu u 2 u d Q~ ~d 2 d s s dy • Thermal diffusivity muDu 1 2 ru 2 u T k ~d ~ s n n0 2 19 The temperature rise from collisions is very small Energy Sink DE ~ (1 e 2 ) 12 mu 2 • energy loss per collision • Energy loss rate per unit volume u DE ~ (1 e ) mn u s 2 1 2 2 • Energy sink coefficient 3 u nT 3/ 2 G~r ~ s n n0 20 Example: Cooling of Granular Gas • Let’s for t=0 T=T0, V=0, n=const T G : g (1 e 2 )T 3/ 2 t • temperature evolution T where g const T0 g (1 e )t 1 1 2 2 2 • asymptotic behavior T ~ 1/t2 • homogeneous cooling is unstable with respect to clustering!!! 21 Q: Does the temperature reach 0 in finite time? R: Difficult to say, in simulations sometimes it does. Clustering Instability Simulations of 40,000 discs, e=0.5 Init. Conditions: uniform distribution Time 500 collisions/per particle MacNamara & Young, Phys. Fluids, 1992 Goldhirsch and Zanetti, PRL, 70, 1619 (1993) Ben-Naim, Chen, Doolen, and S. Redner PRL 83, 4069 (1999) Mechanism of instability: decrease in temperature → decrease in pressure→ increase in density→ increase in number of collisions → increase of dissipation→ decrease in temperature …. 22 Thermo-granular convection • inversed temperature profiles: temperature is lower at open surface due to inelastic collisions • Consideration of convective instability Theory:Khain and Meerson PRE 67, 021306 (2003) Experiment: Wildman, Huntley, and Parker, PRL 86, 3304 (2001) Shaking A=A0 sin(wt) 23 Kinetic Theory • Boltzmann Equation for inelastically colliding spherical particles of radius d f ( v, r, t ) v f ( v, r , t ) B ( f , f ) t • f(v,r,t) – single-particle collision function, 24 Collision integral • binary inelastic collisions • molecular chaos • splitting of correlations: f(v1,v2,r1,r2,t)= f(v1,r1,t) f(v2,r2,t) loss ter m 64444444 47 4444444 4 8 1 B( f , f ) d 2 dv 2 dk (k v12 ) 2 f ( v1 ,r, t ) f ( v2 ,r, t ) f ( v1 ,r, t ) f ( v 2 , r, t ) k v12 0 e 442 44444444443 144444444 gain te r m • k – vector along impact line • v’1,2 –precollisional velocities • v1,2 –postcollisional velocities 25 Macroscopic variables Afdv • averaged quantity A • stress tensor s ij n ui u j • heat flux Q j 12 n u 2u j • energy sink G 1 n (1 e 2 )d 2 8n 3 d v d v v 1 2 12 f ( v1 ) f ( v 2 ) • approximations for f(v,r,t) in Eli’s lecture 26 Expressions for smooth inelastic spheres Copied from Bougie et al, PRE 66, 051301 (2002) • equation of state • shear viscosities • bulk viscosity p nT 1 2(1 e )G( ) h1 nd T 165 G ( ) 54 (1 12 )G (v) 2 6G ( ) h2 8ndG ( ) T 3 Smooth inelestic spheres, from Jenkins & Richman, Arch. Ration. Mech. Anal. 87, 355 (1985). 27 Expressions for smooth inelastic spheres Copied from Bougie et al, PRE 66, 051301 (2002) • heat conductivity • energy sink 5 k nd T 245 G ( ) 56 (1 932 )G( ) 2 8G( ) 8nG( )T 3/ 2 G 1 e 2 d Smooth inelastic spheres, from Jenkins & Richman, Arch. Ration. Mech. Anal. 87, 355 (1985). 28 Radial distribution function • =(/6)nd3-packing fraction • dilute elastic hard disks (Carnahan & Starling) (1 7 /16) G ( ) 2 1 • High densities (~c =0.65 closed-packed density in 3D) G ( ) 1 c 4v 3 c 1 1 : c 29 Asymptotic behaviors Works pretty well for sheared granular flows Dilute shear viscosity Nearly closed packed h1 heat conductance k heat sink G 5 16 d 2 25 32 d 4 3 2 T T n 2 d 2T 3/ 2 1 e 2 h1 : T n0 n T k: n0 n n 2 d 2T 3/ 2 G: 1 e 2 n0 n 30 Comparison with MD: Dynamics of Shocks Q: Why is there not a big temperature gradient? R: There is a slow vibration, fast vibrations have a large temperature gradient J. Bougie, Sung Joon Moon, J. B. Swift, and Harry L. Swinney Phys. Rev. E 66, 051301 (2002) 31 Q: Do these equations predict oscillons, waves, etc? Comment: These equations work well for low density and restitution coefficient near 1. R: Oscillons no, waves and bubbles yes. References • • • • Review: -I. Goldhirsch, Annu. Rev. Fluid Mech 35,267 (2003) Phenomenological Hydrodynamics: -P.K. Haff, J. Fluid Mech 134, 401 (1983) Derivation from kinetic theory: -J. Jenkins and M. Richman, Arch. Ration. Mech. Anal. 87, 355 (1985). -J.T. Jenkins and M.W. Richman, Phys. Fluids 28, 3485 (1985) -N. Sela, I. Goldhirsch, J. Fluid Mech 361, 41 (1998) Comparison with simulations: -J. Bougie, Sung Joon Moon, J. B. Swift, and Harry L. Swinney Phys. Rev. E 66, 051301 (2002) -S. Luding, Phys. Rev. E 63, 042201 (2001) -B. Meerson, T. Pöschel, and Y. Bromberg Phys. Rev. Lett. 91, 024301 (2003) 32