RESOURCE LETTER

Roger H. Stuewer, Editor

School of Physics and Astronomy, 116 Church Street SE, University of Minnesota, Minneapolis, Minnesota 55455 This is one of a series of Resource Letters on different topics intended to guide college physicists, astronomers, and other scientists to some of the literature and other teaching aids that may help improve course content in specied elds. The letter E after an item indicates elementary level or material of general interest to persons becoming informed in the eld. The letter I, for intermediate level, indicates material of somewhat more specialized nature; and the letter A indicates rather specialized or advanced material. No Resource Letter is meant to be exhaustive and complete; in time there may be more than one letter on some of the main subjects of interest. Comments on these materials as well as suggestions for future topics will be welcomed. Please send such communications to Professor Roger H. Stuewer, Editor, AAPT Resource Letters, School of Physics and Astronomy, University of Minnesota, 116 Church Street SE, Minneapolis, MN 55455; e-mail: rstuewer@physics.umn.edu.

Resource Letter GP-1: Granular physics or nonlinear dynamics in a sandbox

James Kakaliosa)
School of Physics and Astronomy, University of Minnesota, Minneapolis, Minnesota 55455

Received 25 June 2003; accepted 3 September 2004 This Resource Letter provides a guide to the literature on the statics and dynamics of granular media. Journal article and book references are provided for the following topics: Packing, Angle of Repose, Avalanches and Granular Flow, Hoppers and Jamming, Vertically Vibrated Induced Phenomena, Avalanche Stratication, and Axial Segregation. 2005 American Association of Physics
Teachers.

DOI: 10.1119/1.1810154

I. INTRODUCTION Within every boulder, a grain of sand lies waiting. Large rocks are lifted to the tops of distant mountains through plate tectonics, where vegetation and erosion break them into smaller and smaller pieces. Gravity pulls them down into valleys, where water runoff from rain or melting snow carries the smaller particles the size distribution can range from meters to ne silt toward the ocean.1 A given grain of sand takes approximately 10 000 years to be transported one mile closer to the ocean. Smaller particles are easier to move, so that the average size of the grains continually decreases in the river currents, until at the oceans edge there is a fairly monodisperse size distribution of sand grains. Granular material extends underwater approximately 100 miles from the watersand interface termed the beach before oceanic pressure compresses the granular material back into solid rock. At that point plate tectonics results in either subduction with consequent melting of the material into its constituent elements, or uplifting during the formation of new mountain ranges. In the latter case the entire cycle may then be repeated. If a complete cycle requires 100 million years, then certain grains of sand on your favorite beach may have undergone several such round trips to bring them to their present state.1,2 A typical grain of sand has a diameter ranging from 1 mm to 400 m. According to Brown and Richards,3 granular media consist of discrete solids in direct physical contact most of the time, as compared to slurries, uidized beds and
a

Electronic mail: Kakalios@umn.edu Am. J. Phys. 73 1, January 2005 http://aapt.org/ajp

suspensions. Granular material of diameter less than 100 m are referred to as powders, while those of diameter less than 10 m are superne powders and those below 1 m are hyperne powders. At the other extreme, any particulate with a diameter greater than 3 mm is a pebble, an aggregate such as employed in concrete or a rock, depending on its size. There is a lot of sand on the planetapproximately 10 million cubic miles, all told. Over 10% of the Earths surface consists of deserts, and if all the sand in the world were deposited on the United States it would cover this country to a thickness three miles high.1 A select group of physicists have investigated the properties of granular media, dating back to the 17th century. Coulomb addressed the issue of the origin of the static angle of repose,77 dened as the maximum angle that a sandpile can be constructed while remaining stable against gravity-driven avalanches. The nature of the mechanisms underlying a sandpiles angle of repose remains a topic of intense research interest today. Once the sandpile begins to ow, the moving sand expands under shear, so that the grains of sand move out of the way of each other. This dilatancy was noted and described by Osborne Reynolds,91 whose contributions to uid dynamics are honored, in part, by the dimensionless number that bears his name. The patterns and structures that wind-driven sand create were catalogued and analyzed by Ralph A. Bagnold,4 whose pioneering investigations of Aeolian geomorphology were similarly recognized and commemorated through a dimensionless number the Bagnold number, expressing the ratio of the forces due to friction and collisions between solids to the forces arising from the surrounding viscous uid. So closely is Bagnold associated
2005 American Association of Physics Teachers 8

with the study of sand dunes that he appears as a minor character in Michael Odjedes 1992 novel The English Patient Alfred A. Knopf Inc., New York. In addition to its intrinsic scientic interest and geological signicance, the study of the statics and dynamics of granular materials has profound industrial and commercial applications. Nearly 80% of everything manufactured or grown in this country exists in at least one stage in its development as a granular material. Expenses involved with powder processing in the pharmaceutical, agricultural, and construction industries in this country alone are estimated to be on the order of $80 billion a year. Approximately 3% of all the electrical power generated in this country is utilized in the single application of grinding metal ores into powders, which are then separated using surfactants before alloying and recasting.33,34 Nearly 20 pounds of surfactants for every man, woman, and child on the planet are manufactured every year, with their largest application being the otation of metal particles from low-yield ores.12 No doubt owing to its broad and immediate relevance for practical applications, the study of granular materials has been effectively ignored by physicists aside from the few noteworthy exceptions mentioned above until fairly recently, and investigations of the properties of granular systems have been primarily carried out by chemical, civil, and mechanical engineers. This situation changed in 1987, with the publication in Physical Review Letters of Self Organized Criticality: A Universal Explanation for 1/ f Noise by Per Bak, Chao Tang, and Kurt Wiesenfeld,92 who presented a mathematical model for uctuations in disordered systems, in particular for their tendency to exhibit spectral densities that vary as the inverse of the frequency f hence the descriptive term 1/ f noise. Bak, Tang, and Wiesenfeld used as their representative model the size distributions of avalanches for a sandpile at the critical angle of repose to which additional sand is added at a uniform rate. Their model predicted that this simple system would, without any external tuning, organize itself into a critical state as in a critical or second-order phase transition whereby the mass distribution of the avalanches would exhibit a 1/ f power spectrum. Computer simulations of one-dimensional cellular automata sandpiles provided support for their model.92 The great excitement engendered in the physics community by the possibility that self-organized criticality was the start of a new theory for complex systems was tempered two years later. By then experimental evidence indicated that uctuations in the magnitude of avalanches in real sandpiles are described by power spectra characterized by a Lorentzian frequency dependence that is, constant for low frequencies and a 1/ f 2 dependence above a characteristic frequency determined by the systems properties, rather than a power-law frequency dependence.93 Notwithstanding the controversy over the avalanche uctuations, the publication of the Self-Organized Criticality SOC model can be considered a seminal event in the study of the physics of granular media. Regardless of whether or not the SOC model provides any insight into 1/ f noise, it was at least partly responsible for the subsequent wealth of attention paid by physicists to granular systems, inspiring experimental and theoretical investigations of the fascinating nonlinear dynamics exhibited by these deceptively simple seeming systems. The SOC paper, referring to 1/ f noise phenomena, stimulated the interest of those physicists primarily interested in the electronic properties of solids. However, at the same time soft condensed matter physicists were also turning
9 Am. J. Phys., Vol. 73, No. 1, January 2005

their attention to granular media, motivated by their interest in pattern formation in dynamically driven systems.122,132 These papers, also published in 1989, clearly demonstrated, independent of the question of the applicability of SOC, that there was interesting physics waiting to be uncovered in the study of granular systems. Despite extensive and thorough investigations, both theoretical and experimental, carried out by the engineering community over the past 50 years, there remain many important unanswered questions in granular media that physicists have begun to address. One of the more striking phenomena exhibited by granular materials is the size or mass segregation of two or more different granular species when dynamically driven. Rather than leading to further mixing, as commonsense might suggest, spontaneous segregation can be observed when mixtures are vertically shaken, rotated in a horizontal cylinder about its long axis, or simply poured into a vertical HeleShaw cell with narrow plate separations. The simple process of pouring a mixture from a discharge hopper turns out to be not so simple, as convective rolls will develop within the hopper, leading to size segregation. The occurrence of jamming of hopper discharges at the orice has been studied as an example of a kinetically driven transition, not unlike the glass transition. The segregation of granular materials that differs by size, shape, mass, density, or surface roughness is a major concern for the pharmaceutical industry, for example, where granular systems need to be well mixed and homogeneous over length scales of a pill diameter or less. The elucidation of the basic mechanisms underlying this phenomenon, in which physicists have played an important role, will lead to reduced costs and improved efciencies in this and other important industries. The following discussion is a selective overview of some of the fascinating characteristics and phenomena found in granular media. The breadth of this eld, coupled with the rapid pace of publications, has led to the regrettable yet inevitable exclusion of many signicant topics. It would be all too easy to expand the length of this Resource Letter into a full monograph. That would be unnecessary, as many excellent books on this subject are already in print, as indicated in Sec. II B. The reader is advised to begin with Jacques Durans Sands, Powder and Grains.5 At just over 200 pages, it provides a highly readable and comprehensive introduction to the physics of granular media. I will consider in this Resource Letter only granular systems for which the interstitial uid between the grains is dry air. In this case, the forces acting upon any given grain include only gravity, electrostatics, and contact forces from neighboring grains. The presence of water between the granular contacts leads to cohesive forces between grains that can alter dramatically both the statics and dynamics of a sandpile. A complete discussion of the inuence of humidity and its attendant cohesiveness on granular dynamics would warrant a separate Resource Letter. Moreover, I will focus on properties of granular media easily accessible through table-top experiments. A comprehensive review of geomorphological phenomena resulting from Aeolian granular dynamics could readily occupy several Resource Letters. This Resource Letter is organized as follows. A listing of general resources available on granular media precedes separate sections devoted to Packing, Angle of Repose, Avalanches and Granular Flow, Hoppers and Jamming, Vertically Vibrated Induced Phenomena, Avalanche Stratication, Axial Segregation, and Granular Media and Trafc. There is
James Kakalios 9

considerable overlap in the physical concepts in some of these sections, and to some readers the delimitation may seem arbitrary. For example, Secs. III and IV discuss issues related to Packing and the Angle of Repose of sandpiles. These two topics have much in common, and the assignment of certain references to one section or the other necessarily involved a subjective judgment call; as did certain references in Secs. V and VI, Avalanches and Granular Flows and Hoppers and Jamming. It is hoped that breaking these topics into shorter sections results in the material being more accessible than coarse-grained divisions of Statics, Dynamics, and Segregation Phenomena. II. RESOURCES A. Journals Annual Review of Fluid Mechanics CHAOS Chemical Engineering Science Europhysics Letters Granular Matter Nature Journal of Applied Physics A Journal of Engineering Mechanics Journal of Fluid Mechanics Journal of Statistical Physics Physica A and D Physical Review Letters Physical Review E (occasionally A and B as well) Physics of Fluids A Powder Technology Reviews of Modern Physics Transactions of the Institute of Chemical Engineering Science B. Books and major compilations
1. A Scientist at the Seashore, James S. Trel Collier Books, New York, 1984. A popular science book for the general public. Chapters 1012 describe phenomena observed at the beach. E 2. Sand, Raymond Siever Scientic American Library, New York, 1988. Combines striking photographs with a scholarly discussion of granular properties. E 3. Principles of Powder Mechanics, R. L. Brown and J. C. Richards Pergamon, Oxford, 1966. An early, classic text. I 4. The Physics of Blown Sand and Desert Dunes, R. A. Bagnold Methuen, London, 1941. A classic text discussing physical mechanisms underlying Aeolian geomorphology. I 5. Sands, Powders, and Grains: An Introduction to the Physics of Granular Materials, Jacques Duran Springer, New York, 1997. An excellent introduction to the eld, clearly and entertainingly written. E,I 6. Physics of Granular Media, edited by D. Bideau and J. Dodds Nova Science, Commack, NY, 1991. Proceedings from one of the rst conferences on granular media following the Self-Organized Criticality paper. A 7. Flows of Granular Materials, S. B. Savage Udine, Italy, 1992. An engaging perspective by one of the elds top researchers. I 8. Statics and Kinematics of Granular Materials, R. M. Nedderman Cambridge U.P., Cambridge, UK, 1992. Primarily concerned with packing and ows through hoppers. A 9. Granular Matter: An Interdisciplinary Approach, edited by Anita Mehta Springer-Verlag, Berlin, 1993. Collection of review articles. I 10. Non-Linearity and Breakdown in Soft Condensed Matter, edited by K. K. Bardhan, B. K. Chakrabarti, and A. Hansen Springer-Verlag, Berlin, 1994. I 10 Am. J. Phys., Vol. 73, No. 1, January 2005

11. Mobile Particulate Systems, edited by E. Guazzelli and L. Oger Kluwer Academic, Dordrecht, 1995. A 12. Fragile Objects, P.-G. deGennes and J. Badoz Springer-Verlag, New York, 1996. An excellent overview of soft-condensed matter systems. E 13. How Nature Works: The Science of Self Organized Criticality, Per Bak Springer-Verlag, Berlin, 1996. A history of Self-Organized Criticality by one of its originators, for a general audience. E 14. Statics of Granular Media, V. V. Sokolovski Pergamon, New York, 1965. An early, geometric view of granular packing. I 15. Self-Organized Criticality, Emergent Complex Behavior in Physical and Biological Systems, H. J. Jensen Cambridge U.P., Cambridge, 1998. Overview of experimental and simulation studies of the Self-Organized Criticality model, including but not restricted to granular media systems. E, I 16. The Pursuit of Perfect Packing, Tomasa Aste and Denis Weaire IOP, New York, 2000. A readable account of the geometrical issues related to granular packing, with many historical anecdotes. E, I 17. Jamming and Rheology: Constrained Dynamics on Microscopic and Macroscopic Scales, edited by A. J. Liu and S. R. Nagel Taylor & Francis, New York, 2001. A collection of articles providing an overview of granular ow, edited by two of the elds top researchers. I

C. Conference proceedings
18. Powders and Grains 93, edited by C. Thorton Balkena, Rotterdam, 1993. A 19. Trafc and Granular Flow, edited by D. Wolf World Scientic, Singapore, 1995. A 20. Powders and Grains 97, edited by R. P. Behringer and J. T. Jenkins Balkena, Rotterdam, 1997. A 21. Statistical Mechanics in Physics and Biology, Mater. Res. Soc. Symp. Proc. Vol. 463 Materials Research Society, Pittsburgh, PA, 1997. A 22. Physics of Dry Granular Media, edited by H. J. Herrmann, S. Luding, and J. P. Hovi, NATO ASI Series E Vol. 350 Kluwer, Amsterdam, 1998. A 23. The Granular State, Mater. Res. Soc. Symp. Proc. Vol. 627 Materials Research Society, Pittsburgh, PA, 2001. A 24. Powders and Grains 01, edited by Yuji Kishino Balkena, Rotterdam, 2001. A

D. Review articles
25. Powder MixingA Literature Survey, M. H. Cooke, D. J. Stephens, and J. Bridgwater, Powder Technol. 15, 120 1976. I 26. The Segregation of Particulate Materials. A Review, J. C. Williams, Powder Technol. 15, 245254 1976. I 27. Computer Simulation of Granular Shear Flows, C. S. Campbell and C. E. Brennen, J. Fluid Mech. 151, 167188 1985. A good introduction of simulations of granular dynamics. I 28. Physics of the Granular State, H. M. Jaeger and S. R. Nagel, Science 255, 15231531 1992. A highly cited and inuential review. E 29. Instabilites in a Sandpile, S. R. Nagel, Rev. Mod. Phys. 64, 321 325 1992. E 30. The Dynamics of Flowing Sand, R. P. Behringer, Nonlinear Sci. Today 3, 115 1993. E 31. The Dynamics of Sand, Anita Mehta and G. C. Barker, Rep. Prog. Phys. 57, 383 416 1994. I 32. What is Shaking in the Sandbox, H. M. Jaeger, J. B. Knight, C.-h. Liu, and S. R. Nagel, MRS Bulletin, 19 5, 2531 1994. E 33. The Legacy of Neglect in the U. S., J. B. Ennis, J. Green, and R. Davis, Chem. Eng. Prog. 90, 32 43 1994. A good overview of the industrial applications and societal perspectives of granular media. E 34. The Importance of Storage, Transfer and Collection, T. M. Knowlton, J. W. Carson, G. E. Klinzing, and W. -C. Yang, Chem. Eng. Prog. 90, 44 54 1994. Another review of industrial applications of granular media. E 35. The Physics of Granular Materials, H. M. Jaeger, S. R. Nagel, and R. P. Behringer, Phys. Today 49 4, 3238 1996. E James Kakalios 10

36. Granular Solids, Liquids and Gases, H. M. Jaeger, S. R. Nagel, and R. P. Behringer, Rev. Mod. Phys. 68, 12591273 1996. A more technical treatment of the topics considered in Ref. 35. I 37. A broad collection of review articles is contained in a special issue of Chaos 9, 3 1999. A 38. Built Upon Sand: Theoretical Ideas Inspired by Granular Flows, Leo P. Kadanoff, Rev. Mod. Phys. 71, 435 444 1999. I 39. Granular Matter: A Tentative View, P. G. de Gennes, Rev. Mod. Phys. 71, S374 S382 1999. I 40. Nonequilibrium Patterns in Granular Mixing and Segregation, T. Shinbrot and F. J. Muzzio, Phys. Today 53 3, 2530 2000. E

E. Web sites
41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. http://www.granular.com http://chaos.ph.utexas.edu/research/granular/granular.html http://mrsec.uchicago.edu/granular http://sol.rutgers.edu/shinbrot/Group Index.html http://summa.physik.hu-berlin.de/kies/ http://www.haverford.edu/physics-astro/Gollub/lab.html http://www.ica1.uni-stuttgart.de/ http://pg.chem-eng.nwu.edu/mixing/ http://widget.ecn.purdue.edu/psl/home.html http://physics.clarku.edu/akudrolli/silo.html http://super.phys.northwestern.edu/pbu/ http://www.phy.duke.edu/bob/ http://www.phys.ntnu.no/fossumj/cpx/index.html http://www.physics.umn.edu/groups/sand

III. PACKING The volume occupied by granular particles in a container is determined by their conguration, and depends on local contact forces, the containers geometry, friction between grains and with the container walls, and the details of the systems history. In this sense granular media have more in common with nonequilibrium melt-quenched glasses than thermodynamic solids or liquids, whose properties are independent of thermal history. A grain of sand or powder is a macroscopic object, with many internal degrees of freedom. Consequently any kinetic energy introduced into the granular system is rapidly dissipated through inelastic collisions between particles. Therefore, in the absence of an external driving force, the conguration of the pile will remain unchanged. For a granular system with dry inert air that is, considering only the static case of no net air ow as the interstitial medium, the intergrain forces are entirely classical, consisting of contact and frictional forces, both being electrostatic in nature. In the absence of cohesive forces, the geometry of a granular system is determined solely by gravity and the boundaries of the container. The gravitational potential energy of a particle in a sandpile is approximately 1012 times greater than the thermal energy kT at room temperature.29,36 One can therefore consider granular media to be thermodynamically at zero degrees, and the properties of the sandpile are determined by the specic details under which the pile was constructed. An important consequence is that within a granular system, stable arches and voids may develop, so that a vertical load owing to a mass of grains can have a signicant horizontal component. A graphic illustration of this phenomenon is reected in the collapse of grain silos, which frequently fail owing to large pressures on their sides rather than at their base.52 These arches play a role in dilatancy,91 in which a vertical force applied to the top of a granular system leads to the creation of new voids, which decreases the systems density. That is, sand deforms in the
11 Am. J. Phys., Vol. 73, No. 1, January 2005

presence of shear stresses, and will only compact under isotropic pressure. This accounts for the dry footprints one leaves behind when walking along a wet, sandy shoreline. The water-saturated sand underfoot expands and develops new voids and pores, into which water drains, leaving the top surface drier than its surroundings. Capillary action will eventually restore the top surface to its prior wet appearance. The possibility that stable arches can form indicates that the volume an ensemble of granular material occupies is a highly sensitive function of the connections between grains. The densest packing of granular material corresponds to a crystalline hexagonal close-packed or face-centered-cubic structure for spherical grains, which has a packing fraction of 0.74. The random close packing of spheres has a packing fraction of 0.63, while a packing fraction of 0.55 is found for a random loose packing in the limit of zero gravitational force, determined by studying glass spheres in a liquid whose density is chosen to approach neutral buoyancy conditions.55,56 The difference between these two packing extremes corresponds to a variation in the average intergrain separation of only 10%. 57,58 A granular system prepared in a random conguration and then disturbed, such as by periodic vertical taps to the container, will explore some portion of congurational space in a limited way as it settles to a lower volume contents may have settled during shipping. The dynamics of a granular medium as it slowly relaxes with a logarithmic time dependence to a denser conguration has been mapped to the random-parking problem.73 Even after 106 separate taps, a granular system in a cylindrical container whose height is very much larger than its diameter shows no evidence of reaching a timeindependent stable nal state.59 63,68 76 The large number of internal degrees of freedom for any given grain of sand results in collisions being highly inelastic. In this way kinetic energy between interacting particles is rapidly lost, and clustering or clumping of particles moving in a restricted geometry is observed. This effect, termed inelastic-collapse, leads to a divergence of the collision frequency for some of the particles. This signicant experimental phenomenon is a major complication in numerical simulations.64 67
55. Random Loose Packings of Uniform Spheres and the Dilatancy Onset, George Y. Onoda and Eric G. Lininger, Phys. Rev. Lett. 64, 27272730 1990. I 56. Random Packings of Spheres and Fluidity Limits of Monodisperse and Bidisperse Suspensions, Andrew P. Shapiro and Ronald F. Probstein, Phys. Rev. Lett. 68, 14221425 1992. A 57. A Model for the Packing of Irregularly Shaped Grains, C. C. Mouneld and S. F. Edwards, Physica A 210, 301316 1994. I 58. Perturbative Theory of the Packing of Mixtures and of Non-Spherical Particles, R. B. S. Oakeshott and S. F. Edwards, Physica A 202, 482 498 1994; The Statistical Mechanics of Granular Systems Composed of Spheres and Elongated Grains, S. F. Edwards and C. C. Mouneld, ibid. 210, 290300 1994. These papers describe an innovative attempt to develop a statistical mechanics for powders. While interactions in a thermodynamic system lead to a decrease of the systems energy, the corresponding variable, these authors suggest, is the volume occupied by a granular material. Theoretical discussions of the variation in the packing fraction as the system is perturbed are discussed. I 59. Density Relaxation in a Vibrated Granular Material, James B. Knight, Christopher G. Fandrich, Chen Ning Lau, Heinrich M. Jaeger, and Sidney R. Nagel, Phys. Rev. E 51, 39573963 1995. Experimental measurements of the time dependence of the volume occupied by monodisperse granular media in a tall narrow cylinder as a function of periodic vibrations. A James Kakalios 11

60. Intermittencies in the Compression Process of a Model Granular Medium, A. Ngadi and J. Rajchenbach, Phys. Rev. Lett. 80, 273276 1998. A 61. Fluctuations in Granular Media, Daniel W. Howell, R. P. Behringer, and C. T. Veje, Chaos 9, 559572 1999. Review of experimental studies and numerical simulations of force chains in granular media slowly sheared by a rotating annulus. These forces are quantied by shearing polymer disks and viewing them through crossed polarizers. The spectral density of the stress uctuations is found to exhibit a power-law frequency dependence A. 62. A Tetris-Like Model for the Compaction of Dry Granular Media, E. Caglioti, V. Loreto, H. J. Herrmann, and M. Nicodemi, Phys. Rev. Lett. 79, 15751578 1997. A theoretical investigation of compaction and packing as grains are added from the top of a pile, using a Tetris-like model to mimic the inuence of varying particle shapes. A 63. Studies of Columns of Beads Under External Vibrations, S. Luding, E. Clement, A. Blumen, J. Rajchenbach, and J. Duran, Phys. Rev. E 49, 1634 1646 1994. Experimental and numerical studies of a onedimensional column of steel beads experiencing external vibrations, as the number of beads, the degree of agitation, and the restitution coefcient are systematically varied. Clustered states are observed for high dissipation and/or large numbers of particles, as well as a transition from a condensed phase to a uidized phase as the acceleration of the base plate is increased. I, A 64. Clustering Instability in Dissipative Gases, I. Goldhirsch and G. Zanetti, Phys. Rev. Lett. 70, 16191622 1993. Analytical calculations and numerical simulations of a gas of inelastically colliding particles, nding that the system is unstable against the formation of high-density clusters. A 65. Inelastic Collapse and Clumping in a One-Dimensional Granular Medium, Sean McNamara and W. R. Young, Phys. Fluids A 4, 496 1992; Inelastic Collapse in Two-Dimensions, Phys. Rev. E 50, R28 R31 1994. Theoretical modeling and molecular dynamic simulations for one- and two-dimensional gases of inelastic disks are described. From an initially random conguration, a nite-time singularity can appear, leading to increasing numbers of collisions and the resulting formation of clusters. A 66. Cluster Formation Due to Collisions in Granular Material, A. Kudrolli, M. Wolpert, and J. P. Gollub, Phys. Rev. Lett. 78, 13831386 1997. Experimental investigation of spherical particles rolling on a smooth surface, driven by an oscillating sidewall. Inelastic collisions lead to the formation of clusters. I 67. Cluster-Growth in Freely Cooling Granular Media, S. Luding and H. J. Herrmann, Chaos 9, 673 681 1999. Numerical simulations of the time dependence of inelastic collapse of hard spheres and subsequent aggregate formation are described. A 68. Uniaxial Compression Effects on 2D Mixtures of Hard and Soft Cylinders, T. Travers, D. Bideau, A. Gervois, J. P. Troadec, and J. C. Messager, J. Appl. Phys. A 19, L1033 1986. An experimental study of the macroscopic stressstrain law for a mixture of hard and soft cylinders examines the role that geometric and compositional heterogeneities play. I 69. Force Fluctuations in Bead Packs, C.-H. Liu, S. R. Nagel, D. A. Schecter, S. N. Coppersmith, S. Majumdar, O. Narayan, and T. A. Witten, Science 269, 513 1995. Experimental studies and numerical simulations of force chains in bead packs are described. The uctuations in the force distribution arise from variations in contact angles between beads along with the constraint of the force balance due to every other bead in the pile. I 70. Contact Forces in a Granular Packing, Farhang Radjai, Stephane Roux, and Jean Jacques Moreau, Chaos 9, 544 550 1999. Numerical simulations of two- and three-dimensional granular packings, conrming the existence of force chains and an exponential distribution of contact forces. A 71. Granule-by-Granule Reconstruction of a Sandpile from X-Ray Microtomography Data, G. T. Seidler, G. Martinez, L. H. Seeley, K. H. Kim, E. A. Behne, S. Zaranek, B. D. Chapman, S. M. Heald, and D. L. Brewe, Phys. Rev. E 62, 8175 8181 2000. A 72. Footprints in Sand: The Response of a Granular Material to Local Perturbations, Junfei Geng, D. Howell, E. Longhi, R. P. Behringer, G. Reydellet, L. Vanel, E. Clement, and S. Luding, Phys. Rev. Lett. 87, 35506 35510 2001. Experimental report of the ensemble-averaged response of granular packings to point forces applied at the top of the pile of polymer discs, whereby the force chains are imaged using 12 Am. J. Phys., Vol. 73, No. 1, January 2005

73.

74.

75.

76.

crossed polarizers. Ordered packings have a long-range propagative force component not found in disordered sandpiles. A Granular Relaxation Under Tapping and the Trafc Problem, D. C. Hong, S. Yue, M. Y. Choi, and Y. W. Kim, Phys. Rev. E 50, 4123 4135 1994. The volume relaxation of a one-dimensional granular column is investigated within the context of a diffusing void model. The volume occupied by a granular system as a function of perturbations is related to the problem of partitioning automobiles in a disordered parking lot. Steady-state traveling wave solutions are found that can then account for the discontinuous stickslip reduction of the columns height under continuous tapping. A Packing of Compressible Granular Materials, Hernan A. Makse, David L. Johnson, and Lawrence M. Schwartz, Phys. Rev. Lett. 84, 4160 4163 2000. Computer simulations are compared to experimental measurements of the packing fraction of three-dimensional granular media as a function of external conning stress, indicating the presence of localized force chains. A Inuence of Shape on Ordering of Granular Systems in Two Dimensions, I. C. Rankenburg and R. J. Zieve, Phys. Rev. E 63, 61303 61312 2001. Experiments and simulations are reported for twodimensional granular materials consisting of various noncircular shapes. A Random Packings of Frictionless Particles, C. S. OHern, S. A. Langer, A. J. Liu, and S. R. Nagel, Phys. Rev. Lett. 88, 9570395707 2002. Numerical simulations of random ensembles of frictionless granular media, intended to model observed exponential distribution of forces within pile. A

IV. ANGLE OF REPOSE One of the earliest scientic studies of granular media addressed the basic physical mechanisms underlying the angle of repose of a dry, cohesiveless sandpile.77 When individual grains of sand are dropped from a given vertical height onto a horizontal surface, they come to rest through inelastic collisions with the horizontal base plate and with each other. If a grain lands on top of another grain, it will most likely tumble off, unless the surrounding area is already occupied with other sand grains. In this way the dropped grains form a sandpile. If the location of the source of falling grains does not vary, the resulting pile will be approximately conical in structure, with a triangular vertical cross section where the free surface makes an angle m with the horizontal. This is termed the maximum angle of stability. For a sandpile whose free surface makes an angle shallower than m with the horizontal, the random conguration of the sand grains held together by contact normal forces, intergrain friction, and gravity will be stable excepting the possibility of frictional creep101. The pile will remain in this conguration indenitely unless the base plate is disturbed or further grains are added to the pile. In this sense the sandpile can be considered a solid, for no uid, regardless of its viscosity, will retain its shape against gravity for sufciently long times. If the base plate is now tilted, so that the free surface of the conical pile makes an angle larger than m with the horizontal, gravitational forces overwhelm frictional drag and the sand grains on the surface of the pile form an avalanche down the surface.78,79 At this point the sandpile pours like a uid, though with a crucial distinction from Newtonian uids discussed in Sec. V. This avalanche continues until the free surface makes a smaller angle r , termed the angle of repose, with the horizontal surface. Experimental and numerical investigations have found that the angle of repose r for a dry sandpile depends on the density of the grains, the coefcient of sliding and rolling friction, particle size, and surface roughness and shape.80 86 The presence of a small amount of interstitial liquid introduces cohesive forces owing
James Kakalios 12

Fig. 1. Image of a two-dimensional pile of polymer discs whose axis of polarization rotates under pressure when viewed between crossed polarizers. The bright regions indicate discs under compressive strain, indicating the force chains in the pile Reprinted with permission from Ref. 83. Copyright 2001 by the A.P.S.

to the liquid bridges between neighboring particles. This cohesive force can overwhelm gravity and intergrain friction, and the free surface of a moist sandpile can be nearly perpendicular to the horizontal, a benecial property during the construction of sandcastles at the beach. The distribution of applied loads is highly nonuniform for a random geometry of contacts in a dry, cohesionless sandpile. The inhomogeneous network of contacts leads to the development of force chains, through which loads percolate within the sandpile. In this way regions within a sandpile may be shielded from forces, whether an externally applied load or the weight of the material above it. Detailed experimental investigations have explored the inuence of intergrain friction and packing order on the length and distribution of these force chains. The packing-order term refers to the detailed history of the grains of sand within the pile. Two sandpiles, containing the same number of the same type of particle but differing in their placement, either in a crystalline conguration or an amorphous pile, will differ greatly in their force network distribution.80,83,87,88 These force chains can be imaged in two-dimensional systems using polymer discs whose polarization axis rotates under strain. As shown in Fig. 1, when a two-dimensional sandpile is viewed through crossed polarizers, the weight of the pile is selectively carried by a subset of random contacts which appear bright in this image throughout the system.52,83 The sensitivity of the sandpiles mechanical properties on the intergrain connections is also reected in its ability to propagate sound.89,90 Acoustic vibrations depend critically on the compression and expansion of contacts between neighboring grains. Consequently, in the long-wavelength limit long, that is, compared to both the diameter of the granular species and any arches the variation of contact pressure with depth in the granular system leads to a nonlinear dispersion of acoustic waves, the net effect being to deect horizontal sound waves orthogonally to the direction of propagation. Owing to this acoustical dispersion, granular media function as highly effective vibration-isolation systems. Time-of-ight experiments nd that the leading edge of an acoustical pulse in granular material corresponds to a speed of sound of 280 m/s. In addition, at lower wavelengths, the acoustic response displays a noisy frequency dependence that is reproducible and stable in time. However, if the granular system is disturbed such as by running ones nger through the sandbox the noisiness of the frequency response changes to another, stable, reproducible pattern. Clearly, the detailed nature of the force chains throughout the granular system plays a central role in determining the properties of the fundamental excitations of the system.
13 Am. J. Phys., Vol. 73, No. 1, January 2005

77. Essay on the Rules of Maximis and Minimis Applied to Some Problems of Equilibrium Related to Architecture, C. A. Coulomb, Acad. R. Mem. Phys. Divers Savants 7, 343 1773. Frequently cited and seldom read, historically signicant as one of the rst scientic papers on the physical mechanisms underlying granular packing. A 78. Stochastic Model for the Motion of a Particle on an Inclined Rough Plane and the Onset of Viscous Friction, G. G. Batrouni, S. Dippel, and L. Samson, Phys. Rev. E 53, 6496 6503 1996. A stochastic model is developed of particles moving down a roughened inclined plane that nds that the frictional force is proportional to the velocity rather than the expected square of the velocity. These results are in agreement with experimental measurements. A 79. Experiments on a Gravity-Free Dispersion of Large Solid Spheres in a Newtonian Fluid Under Shear, R. A. Bagnold, Proc. R. Soc. London, Ser. A 225, 49 63 1954. Of early, historical signicance. E 80. Force Distributions in Three-Dimensional Granular Assemblies: Effects of Packing Order and Interparticle Friction, D. L. Blair, N. W. Mueggenburg, A. H. Marshall, H. M. Jaeger, and S. R. Nagel, Phys. Rev. E 63, 41304 41311 2001. Experimental measurements of the normal force distribution for a vertical container with a triangular cross section are reported. The force-distribution function is essentially identical for random and hexagonal close-packed crystalline arrangements. I 81. Interfacial Friction of Powders on Concave Counterfaces, B. J. Briscoe, L. Pope, and M. J. Adams, Powder Technol. 37, 169181 1984. An experimental study of frictional forces, dynamic angle of repose, and normal forces at the container walls of granular media in a rotating cylinder are reported as a function of particle mass and rotational velocity. I 82. Friction in Granular Flows, H. M. Jaeger, Chu-Heng Liu, S. R. Nagel, and T. A. Witten, Europhys. Lett. 11, 619 624 1990. Early attempt to develop analytical model for angle of repose. E 83. Memory in Two-Dimensional Heap Experiments, Junfei Geng, Emily Longhi, R. P. Behringer, and D. W. Howell, Phys. Rev. E 64, 60301 60304 2001. Measurements of the force-chain distribution in a two-dimensional sandpile using photoelastic discs viewed through crossed polarizers see also Ref. 52, as a function of the sandpileconstruction history. A 84. Grain Non-Sphericity Effects on the Angle of Repose of Granular Material, Jason A. C. Gallas and Stefan Sokolowski, Int. J. Mod. Phys. B 7, 20372046 1993. Simulations of two-dimensional sandpiles, taking into account that real sand grains are not perfect spheres. A 85. Numercial Investigation of the Angle of Repose of Monosized Spheres, Y. C. Zhou, B. H. Xu, and A. B. Yu, Phys. Rev. E 64, 2130121309 2001. Simulations, using discrete element method, of the angle of repose, investigating sensitivity to various parameters including coefcient of sliding and rolling friction, particle size, and container thickness A. 86. Effect of Grain Anisotropy on Ordering, Stability and Dynamics in Granular Systems, C. J. Olson, C. Reichhardt, M. McCloskey, and R. J. Zieve, Europhys. Lett. 57, 904 2002. Simulations and experiments investigating the role of grain-aspect ratio on the angle of repose in two-dimensional sandpiles. A 87. Signatures of Granular Microstructure in Dense Shear Flows, D. M. Mueth, G. F. Debregeas, G. S. Karczmar, P. J. Eng, S. R. Nagel, and H. M. Jaeger, Nature London 406, 385389 2000. Noninvasive measurements are employed to characterize the steady-state shear ow of granular media in a three-dimensional Couette geometry. The shape of the velocity prole is characterized by two length scales independent of the height and shear rate, but is sensitive to the grains morphology. A 88. Stresses Developed by Dry Cohesionless Granular Materials Sheared in an Annular Shear Cell, S. B. Savage and M. Sayed, J. Fluid Mech. 142, 391 430 1982. Experimental studies of granular material in Couette-geometry shear cells. I 89. Sound in Sand, Chuheng Liu and Sidney R. Nagel, Phys. Rev. Lett. 68, 23012304 1992. Measurements of the spectral density of vibrations recorded through a three-dimensional granular system are reported. The power spectra decay with a power-law frequency dependence over ve decades in frequency. The stability of the frequency response was an early indication of the inuence of force chains on the granular systems dynamical behavior. A 90. Spatial Patterns of Sound Propagation in Sand, Chu-heng Liu, Phys. James Kakalios 13

Fig. 2. Time-lapse image of a pile of mustard seeds, tilted to an angle greater than the angle of maximum stability. The sharp boundary between avalanching seeds and those maintaining a rigid pile is evident Reprinted with permission from Ref. 36. Copyright 1996 by the A.P.S.

Rev. B 50, 782794 1994. Experimental investigation of the fundamental acoustic excitations of granular media. Time-of-ight measurements are described characterizing the speed of sound in granular media. I, A

V. AVALANCHES AND GRANULAR FLOW As mentioned in Sec. I, physical sandpiles do not display the 1/ f distribution in avalanche size that the Self-Organized Criticality model92 was intended to account for, using cellular automata simulations of sandpiles as the archetypal example. Initially, there was some controversy over the Chicago results, which found that the distribution of avalanches had a Lorentzian frequency dependence rather than a powerlaw frequency dependence.93 A Lorentzian power spectrum for the distribution of avalanches indicates that there is a characteristic size to the granular avalanches. This agrees with the observation that only grains within a wedge between the maximum angle of stability m and the angle of repose r participate in an avalanche in a continuously driven system. However, Held and co-workers at I.B.M. in 1990 observed avalanche distributions consistent with SOC predictions.94 After a dispute over whether the ndings depended on the use of Chicago sand or New York sand, a consensus developed that Helds results were due to nite-size effects, and when the size of the experimental system is increased, the agreement with SOC disappears.94 97 A recent development is the report of avalanche distributions as predicted by Self-Organized Criticality when the granular system consists of long-grain uncooked rice, but not when spherical beads or sand grains are employed. The role that the highly asymmetric geometry of the granular media may play in the avalanche dynamics remains under investigation.98,99 One of the complications in developing a hydrodynamic model for granular materials is that the ow of powders or grains is strongly non-Newtonian.100109 For example, when a static sandpile at the angle of marginal stability is further tilted, the resulting movement, driven by gravity, is restricted to a narrow region typically ten grains down from the free surface, as shown in Fig. 2.36 This is in marked contrast to the manner in which a Newtionian uid, such as water, would ow. In that case only the uid in direct contact with the rigid base plate would be stationary forming the no14 Am. J. Phys., Vol. 73, No. 1, January 2005

slip zone and the velocity of the water increases with distance from the base plate. For the granular system, the entire pile, save the top surface, is immobile, and there is a fairly sharp boundary, roughly one or two grains wide, separating the owing region from the underlying static, rigid pile. Magnetic-resonance-imaging studies of continuously avalanching grains in a rotating cylinder nd that the particles velocity increases quadratically with distance from the boundary separating the owing and static grains.110 This velocity discontinuity yields a large shear force at this interface. As in the case of uniaxial compression, sand expands owing to shear, and this ow-driven dilatancy can lead to sieving and segregation of avalanching granular material owing to differing size or density. Recent studies have found that particles below this interface are not perfectly static, but rather exhibit creep, that is, slow motion observed on long time scales, which can extend to arbitrary depth within the sandpile.101 The sensitivity to external driving, coupled with a nonlinear dependence on external constraints and sample history, has complicated efforts to develop a continuum model for granular dynamics.111114 One of the more successful attempts to model granular ow analytically is by Bouchaud and co-authors,113 who represent a sandpile as consisting of two constituents, sticking or rolling grains. The latter are described by a mean velocity with a dispersion factor. The two hydrodynamical variables in the model are the height of the sandpile that is, the density of immobile grains and the density of rolling grains. By allowing for the interconversion between sticking and rolling states, they obtain a hysteresis whereby an angle of maximum stability, greater than the angle of repose, must be exceeded to induce an avalanche of owing sand.
91. On the Dilatancy of Media Composed of Rigid Particles in Contact, O. Reynolds, Philos Mag. 20, 469 1885. Of historical signicance. A 92. Self-Organized Criticality: An Explanation of the 1/ f Noise, P. Bak, C. Tang, and K. Wiesenfeld, Phys. Rev. Lett. 59, 381384 1987. Describes cellular-automata simulations of distributions of avalanches from a continually driven sandpile at the angle of maximum stability. I 93. Relaxation at the Angle of Repose, H. M. Jaeger, C.-h. Liu, and S. R. Nagel, Phys. Rev. Lett. 62, 40 43 1989. Experimental results which contradict the predictions of Self-Organized Criticality model in Ref. 92. I 94. Experimental Study of Critical-Mass Fluctuations in an Evolving Sandpile, G. A. Held, D. H. Solina II, D. T. Keane, W. J. Haag, P. M. Horn, and G. Grinstein, Phys. Rev. Lett. 65, 11201123 1990. Experimental test of the Self-Organized Criticality model, which agrees with the predictions of Ref. 92. I 95. Finite-Size Effects in a Sandpile, C.-h. Liu, H. M. Jaeger, and S. R. Nagel, Phys. Rev. A 43, 70917092 1991. Experimental investigation of the discrepancy between Refs. 93 and 94, attributing the discrepancy to nite-size effects in Ref. 94. I 96. Tracer Dispersion in a Self-Organized Critical System, Kim Christensen, A lvaro Corral, Vidar Frette, Jens Feder, and Torstein Jssang, Phys. Rev. Lett. 77, 107110 1996. Experimental report that avalanches of long-grained rice agree with Self-Organized Criticality model, while similar studies with spherical grains do not. A 97. Avalanches in One-Dimensional Piles with Different Types of Bases, E. Altshuler, O. Ramos, C. Martinez, L. E. Flores, and C. Noda, Phys. Rev. Lett. 86, 54905493 2001. Experimental investigation examining whether the discrepancy between granular avalanches that agree with the SOC model and those that do not can be attributed to a sensitivity to the properties of the baseplate on which the pile resides. The authors conclude that the proper choice of a baseplate can indeed lead to scaling properties of granular avalanches that agree with SOC predictions. A James Kakalios 14

98. Dynamics of a Grain on a Sandpile Model, L. Quartier, B. Andreotti, S. Douady, and A. Daer, Phys. Rev. E 62, 8299 8307 2000. Experimental and theoretical investigations of the motion of a grain moving down an inclined plane comprised of identical grains, wherein the transient and steady-state motions are examined. A 99. Avalanche Dynamics in a Pile of Rice, V. Frette, K. Christensen, A. Malthe-Srenssen, J. Feder, T. Jssang, and P. Meakin, Nature London 379, 4952 1996. Experimental studies of the size distribution of avalanches for a pile at the angle of maximum stability, when the material consists of long-grained rice. The observed avalanche distribution agrees with predictions of the Self-Organized Criticality model. I 100. Grain Flows as a Fluid Mechanical Phenomenon, P. K. Haff, J. Fluid Mech. 134, 401 430 1983. An important, early attempt to describe granular ow using the phenomenology of hydrodynamics. E, I 101. Creep Motion in a Granular Pile Exhibiting Steady Surface Flow, Teruhisa S. Komatsu, Shio Inagaki, Naoko Nakagawa, and Satoru Nasuno, Phys. Rev. Lett. 86, 17571760 2001. Experimental studies that clearly demonstrate that even below the angle of maximum stability, slow relaxation and granular ow occur. I 102. Role of Surface Diffusion as a Mixing Mechanism in a BarrelMixer: Part One, B. H. Kaye and D. B. Sparrow, Ind. Chem. 40, 200205 1964. Experimental study of the curvature of the prole of owing granular particles in a horizontal rotating cylinder as a function of rotation speed. Surface diffusion is found to extend a distance of eight or nine particles beneath the free surface. I 103. Flow Regimes in Fine Cohesive Powders, A. Castellanos, J. M. Valverde, A. T. Perez, A. Ramos, and P. Keith Watson, Phys. Rev. Lett. 82, 1156 1159 1999. Experimental studies of ne, cohesive grains in a horizontal rotating drum are employed to map out the dynamical phase diagram for the transition from solid-like to uidized behavior as a function of velocity and particle size. A 104. Sensitivity of Granular Surface Flows to Preparation, A. Daerr and S. Douady, Europhys. Lett. 47, 324 330 1999. Experimental studies of transient avalanches arising when a cylindrical sandpile collapses under its own weight to form a conical structure. A 105. Scales and Kinetics of Granular Flows, I. Goldhirsch, Chaos 9, 659 672 1999. A review of theoretical attempts to understand clustering and inelastic collapse through kinetic theory for the ow of dilute granular media. A 106. Rapid Gravity Flow of Cohesionless Granular Materials Down Inclined Chutes, M. Sayed and S. B. Savage, J. Appl. Math. Phys. 34, 84 100 1983. A 107. Dynamics of Grain Avalanches, J. Rajchenbach, Phys. Rev. Lett. 88, 1430114304 2002. Experimental study of the nucleation and growth of avalanches of noncohesive grains in a two-dimensional rotating cylinder. The downward front of the avalanche is found to propagate with a velocity approximately twice the average avalanche velocity, while the upper front ows with a speed roughly equal to the average avalanche speed. A 108. Flow in Powders: From Discrete Avalanches to Continuous Regime, J. Rajchenbach, Phys. Rev. Lett. 65, 22212224 1990. Excellent paper investigating the conditions at which discrete, intermittent avalanches transform into a continuous ow. The transformation is found to be hysteretic. I 109. Evolution of a Sandpile in a Thick-Flow Regime, S. N. Dorogovtsev and J. F. F. Mendes, Phys. Rev. E 61, 29092919 2000. Analytic calculations of the evolution of a one-dimensional sandpile, when the thickness of the avalanching layer is large compared to the grain size, for a uniform input ow. A 110. Flow Measurements by NMR, A. Caprihan and E. Fukushima, Phys. Rep. 198, 195235 1990; Nuclear Magnetic Resonance as a Tool to Study Flow, E. Fukushima, Annu. Rev. Fluid Mech. 31, 95123 1999. Comprehensive reviews of the theory and applications of pulsed nuclear-magnetic resonance as a noninvasive probe of the velocity distributions of liquid ow. Extension of this technique to granular particles that contain a liquid center where the protons in the uid are able to follow the rapidly changing magnetic-eld gradients enables determinations of the position and velocity of owing granular media. E, I 111. A Theory of Rapid Flow of Identical, Smooth, Nearly Elastic Particles, J. T. Jenkins and S. B. Savage, J. Fluid Mech. 130, 187202 1983. A 15 Am. J. Phys., Vol. 73, No. 1, January 2005

112. Grain Dynamics in a Two-Dimensional Granular Flow, S. Horluck and P. Dimon, Phys. Rev. E 63, 3130131317 2001. Experimental studies of individual spheres in a granular ow in a small-angle twodimensional funnel, with particular attention to the mechanisms underlying the creation and interaction of shock waves. I 113. A Model for the Dynamics of Sandpile Surfaces, J.-P. Bouchaud, M. E. Cates, J. Ravi Prakash, and S. F. Edwards, J. Phys. I. 4 10 13831410 1994; Hysteresis and Metastability in a Continuum Sandpile Model, Phys. Rev. Lett. 74, 19821985 1995. A continuum model is described for sandpile surfaces. By including two distinct phases of grains; trapped and mobile; and allowing for their interconversion, a realistic description of the observed angles of maximum stability and repose are obtained. A 114. Surface Flows of Granular Mixtures. I. General Principles and Minimal Model, T. Boutreux and P. G. de Gennes, J. Phys. I 6, 1295 1307 1996. The model of Ref. 105 is extended to include mixtures of differing granular media, leading to a continuum description of segregation of the avalanching material. A

VI. HOPPERS AND JAMMING The combination of inhomogeneous force chains in a granular system and non-Newtonian uid dynamics lead to signicant properties when granular media drain gravitationally through an orice in the bottom of a hopper. While the grains ow through the orice as would a uid, the detailed nature of the behavior of the granular material is quite different from that of a Newtonian liquid. For example, a container lled with water, open to atmospheric pressure at the top, will initially discharge rapidly through an opening in the baseplate owing to the large pressure of the volume of water in the tank. As this pressure head drops, the rate of discharge of the water decreases. In contrast, to rst order, the ow of grains of sand through an equivalent open orice occurs at the same rate regardless of the quantity of sand above the opening.8,115121 This is because of the arching and force chains mentioned in Sec. III. The irregular contacts between grains leads to a fraction of the weight of the volume of sand above the orice being transferred horizontally to the sidewalls of the container, rather than hydrodynamically to the material at the opening. The pressure on the grains of sand near the orice arises only from a restricted, hemispherical volume approximately ten grains in radius from the opening, and is roughly independent of the amount of additional material in the container. This region is referred to as the freefall arch, which forms the boundary separating particles that are not in direct contact and accelerate freely owing to gravity, and the packed bed of compressed particles above.30,122124 The mass-ow rate for granular material in a hopper with an orice of diameter D o is observed to vary as 1/2 2 1/2 g 1/2D 5/2 for o as compared to a discharge rate of g D o H liquids where H is the height of the uid in the container.8 Because the discharge rate of granular systems is independent of the amount of material in the container, we have the practical application of the hour glass. When half of the sand has owed through an egg timer, one equates this with half of the time necessary to fully drain the timer, which would only be true if the top container discharged at a constant rate. Another indication that one is not dealing with a Newtonian uid when considering the draining of granular material from a hopper is the evolution of the top surface of granular media. For example, the top surface of grains in a cylindrical hopper with a circular orice in its baseplate initially will maintain a horizontal prole. As the amount of granular material in the cylinder decreases, a conical indentation forms
James Kakalios 15

as the material directly above the orice feels its inuence, while those grains off to the side remain stationary. The exact shape of this indentation along with the timing of its development is sensitive to both the container and orice geometries. At the shear interface between the rigid granular material within the hopper and the owing material draining through the bottom opening there can be signicant convection rolls, which can lead to size segregation of a binary mixture of different size or mass granular materials.8,125,126 An initially homogeneous mixture in a hopper will remain mixed as it empties from a hopper when the orice is rst opened, but will alternate between large and small particles for example as the hopper continues to drain. An understanding of this segregation phenomenon as a mixture empties from a hopper is obviously of importance for many industrial applications, from agricultural to pharmaceutical. The suggestion that granular systems might be considered as analogous to melt-quenched glasses whose properties are highly sensitive to the materials prior thermal history has been extended by Liu and Nagel.127 These authors suggest that the jammed state of a granular material is akin to a frozen glass at low temperatures. Just as raising the temperature of a glass leads to unjamming of the thermally arrested material, lowering the density of a jammed granular system similarly leads to uid-like behavior. While both glasses and sand jam at high densities, the role of temperature in a glass is lled by an externally applied load for a sandpile. Further work is needed to determine whether this conceptual framework can serve as the basis of a more complete theory of granular media.128 130
115. Flow of Granular Material Through Horizontal Apertures,A. Harmens, Chem. Eng. Sci. 18, 297306 1963. Analytical calculation of the mass ow rate of granular matter driven by gravity through a horizontal orice. I 116. The Flow of Granular Materials. I. Discharge Rates from Hoppers, R. M. Nedderman, U. Tuzun, S. B. Savage, and G. T. Houlsby, Chem. Eng. Sci. 37 11, 15971609 1982. The rst of three review papers on the ow of granular media through hoppers, with emphasis on the inuence of the container and orice geometry. I 117. The Flow of Granular Solids Through Orices, W. A. Beverloo, H. A. Leniger, and J. van de Velde, Chem. Eng. Sci. 15, 260269 1961. Experimental investigation of the discharging from a hopper by differing seeds, with the development of a phenomenological expression for the discharge rate. A 118. The Hour-Glass Theory of Hopper Flow, J. F. Davidson and R. M. Nedderman, Trans. Inst. Chem. Eng. 51, 2935 1973. Expressions for the mass-ow rate and stress distributions are presented; based upon experimental investigations of cohesionless granular media owing from a smooth-walled conical hopper. A 119. Funnel Flows in Hoppers, T. V. Nguyen, C. E. Brennen, and R. H. Sabersky, J. Appl. Mech. 47, 729 1980. A 120. Shocks in Sand Flowing in a Silo, A. Samandi, L. Mahadevan, and A. Kudrolli, J. Fluid Mech. 452, 293301 2002. An experimental study of the granular dynamics on the top surface of material emptying from a quasi-two-dimensional silo. I 121. Rate of Discharge of Granular Materials from Bins and Hoppers, H. E. Rose and T. Tanaka, Engineer 208, 465 469 1959. Early, careful study of discharge rates as the particle diameter and coefcient of friction are systematically varied. A 122. Pattern Formation in Flowing Sand, G. William Baxter, R. P. Behringer, Timothy Fagert, and G. Allan Johnson, Phys. Rev. Lett. 62, 28252828 1989. X-ray transmission studies of granular material draining owing to gravity from a conical hopper, demonstrating that the density waves that propagate during discharge are sensitive to the granular materials surface roughness. I 123. Experimental Test of Time Scales in Flowing Sand, G. W. Baxter, R. Leone, and R. P. Behringer, Europhys. Lett. 21, 569574 1993. Power spectrum analysis of granular discharge. I 16 Am. J. Phys., Vol. 73, No. 1, January 2005

124. Granular Flow: Friction and the Dilatancy Transition, Peter A. Thompson and Gary S. Grest, Phys. Rev. Lett. 67, 17511754 1991. Molecular dynamics simulations of two-dimensional granular systems under shear are described. A 125. The Discharge of Fine Sands from Conical Hoppers, T. M. Verghese and R. M. Nedderman, Chem. Eng. Sci. 50, 31433153 1995. A 126. Effects of Horizontal Vibration on Hopper Flows of Granular Materials, M. L. Hunt, R. C. Weathers, A. T. Lee, and C. E. Brennen, Phys. Fluids 11, 68 75 1999. Modications to the mass-ow discharge-rate expression for quasi-two-dimensional hoppers are investigated as periodic horizontal vibrations are applied. I 127. Jamming is Not Just Cool Any More, A. J. Liu and S. R. Nagel, Nature London 396, 2122 1998. Presentation of a potentially signicant model likening a jamming transition in granular media to a glass transition in a thermally arrested liquid. E 128. Force Distributions Near the Jamming and Glass Transitions, C. S. OHern, S. A. Langer, A. J. Liu, and S. R. Nagel, Phys. Rev. Lett. 86, 111114 2001. Experimental and theoretical determinations of interparticle normal-force distributions near the jamming transition. A 129. Jamming of Granular Flow in a Two-Dimensional Hopper, Kiwing To, Pik-Yin Lai, and H. K. Pak, Phys. Rev. Lett. 86, 7174 2001. The jamming probability in a quasi-two-dimensional hopper is measured, and the arch at the orice is described using the formalism of a self-avoiding random walker. A 130. Jamming Phase Diagram for Attractive Particles, V. Trappe, V. Prasad, L. Cipelletti, P. N. Segre, and D. A. Weitz, Nature London 411, 772775 2001. An experimental examination of the solid liquid transition for weakly attractive colloidal particles that indicates support for the model of Liu and Nagel Ref. 127. I

VII. VERTICALLY VIBRATED INDUCED PHENOMENA One consequence of the highly inelastic collisions between grains in a granular system is that any uid-like behavior is only observed when the system is continuously dynamically driven. As mentioned in Sec. II, the conguration of a granular system is determined in large part by the boundary conditions of the container that constrains the particles. One convenient technique for adding external energy to a granular system is through oscillations of the baseplate of the container in either the vertical orthogonal to the plane of the baseplate or horizontal within the plane of the baseplate directions. Michael Faraday reported in 1831 that when a granular system is sinusoidally, vertically vibrated through its base plate, convective rolls and heaping are observed.131 For large amplitude A and high-frequency oscillations , where the acceleration of the container 2 A is larger than the acceleration owing to gravity g , the granular material achieves lift-off from the bottom of the container during certain phases of the vertical shaking cycle. The granular medium then dilates and generates a large-scale convective roll that transports granular material upwards in the center of the system and downwards at the sidewalls in rectangular or cylindrical containers. The net effect is the formation of a stable heap in the center.132137 A change of the containers boundary conditions can reverse the direction of the convective rolls, so that an indentation occurs in the center.134 Grains are continuously owing along the free surface of this heap. While there is no question that this heaping phenomenon results from convective rolls, as conrmed by Magnetic-Resonance-Imaging MRI studies,138 the mechanism by which these rolls are created remains an open question. Friction at the sidewalls and intergrain pressure from the interstitial gas play a role in heap formation as well.153 When an identical system is subjected to horizontal periodic vibrations of the baseplate, convective patterns also have
James Kakalios 16

Fig. 4. Image of a localized vibration-induced excitation, termed an oscillon Reprinted with permission from Ref. 141.

Fig. 3. Patterns observed for a cylindrical cell lled with 100 glass particles, whose baseplate is sinusoidally vibrated, as the vibration frequency and amplitude are varied Reprinted with permission from Ref. 140. Copyright 1994 by the A.P.S.

been observed. Though not as extensively investigated as vertical vibrations, the smoothness of the boundaries is crucial for determining the details of the observed patterns in horizontally shaken systems.139 For vertical vibrations in a large, shallow cylinder containing granular material, the free surface can exhibit a rich collection of standing-wave patterns as the oscillation amplitude and frequency are varied. The similarities to Faraday patterns in vibrated uids is striking in certain circumstances, as shown in Fig. 3, though the physical mechanisms underlying these surface structures is very different in the granular system.140 Umbanhowar, Melo, and Swinney reported141 that for a narrow but denite range of vibration amplitude and frequency, stable, two-dimensional localized excitations in a vibrating layer of sand could be observed. These excitations, which they termed oscillons see Fig. 4, have a propensity to assemble into molecular and crystalline structures. That is, a pair of oscillons beating 180 degrees out of phase with each other, can form a bound dimer pair, similar to a vortexanti-vortex bound pair. Trimers and more complicated structures have also been reported. The fascinating zoology of patterns obtained from vibrating granular materials is not close to being explored exhaustively.141144 A striking segregation effect, whereby large particles in a mixture of granular media of differing sizes rise to the top of
17 Am. J. Phys., Vol. 73, No. 1, January 2005

a cylinder when continuously, periodically shaken in the vertical direction, is known as The Brazil Nut Problem. 145,146 This whimsical title reects the observation that in a large can of mixed nuts, the bigger, heavier Brazil nuts will be found at the top of the mixture, rather than having settled to the bottom of the container. This phenomenon was rst reported when studying the separation of pharmaceuticals as a function of vertical shaking in an effort to mix a disperse system.145 The rising of larger particles to the top of a granular mixture has been attributed to a ratcheting mechanism. That is, an upward uctuation of a large particle induced by the vertical shaking results in a void directly beneath this particle.133,136,146 Other larger particles cannot t into this narrow, constrained space, but smaller particles in the system can. In this way the large particle is prevented from moving downward with subsequent vertical shakes, and the large particle eventually creeps up to the top of the container. An alternative explanation proposes that convective rolls, whose presence is conrmed by MRI measurements, are continuously present as the system is vertically shaken.134,138,147 These convective rolls entrain the larger particles, bringing them to the top surface. Once at the top of the container, the larger particles are unable to follow the convective rolls of the smaller grains that move downward at the sides of the container, and upward in its center. Strong support for the inuence of convective rolls is the observation that alteration of the boundary conditions of the container can cause the rolls to reverse direction. Once the rolls move upward at the sidewalls and downward in the center, larger particles move to the bottom of the container with vertical shaking, rather than to the top. However, the subject is not closed, and experimental and theoretical studies on this question continue.137,148 153
James Kakalios 17

131. On a Peculiar Class of Acoustical Figures; and on Certain Forms Assumed by Groups of Particles upon Vibrating Elastic Surfaces, M. Faraday, Philos. Trans. R. Soc. London, 52, 299340 1831. Historically signicant. A 132. Instability in a Sand Heap, P. Evesque and J. Rajchenbach, Phys. Rev. Lett. 62, 44 46 1989. Early, important experimental study of instabilities in free-surface prole that arise in granular system vertically vibrated. E 133. Vibrated Powders: A Microscopic Approach, Anita Mehta and C. C. Barker, Phys. Rev. Lett. 67, 394 397 1991. A three-dimensional microscopic model is described for a granular system subjected to vertical vibration, with attention paid to the competition between collective and single-particle excitations. A 134. Effects of Container Geometry on Granular Convection, E. L. Grossman, Phys. Rev. E 56, 32903300 1997. A 135. Bubbling in Vertically Vibrated Granular Materials, H. K. Pak and R. P. Behringer, Nature London 371, 231233 1994; Effects of Ambient Gases on Granular Materials Under Vertical Vibrations, H. K. Pak, E. v. Doorn, and R. P. Behringer, Phys. Rev. Lett. 74, 4643 4646 1995. Experimental studies that nd that convection and heaping are sensitive to bubbling of the interstitial gas trapped within the granular media. I 136. Computer Simulation of the Mechanical Sorting of Grains, P. K. Haff and B. T. Werner, Powder Technol. 48, 239245 1986. I 137. Experimental Study of Heaping in a Two-Dimensional Sandpile, E. Clement, J. Duran, and J. Rajchenbach, Phys. Rev. Lett. 69, 1189 1192 1992. Experimental studies of heaping in two-dimensional vertically vibrated systems indicate that a block-slip mechanism at the container walls is responsible for heaping under certain conditions. I. 138. Granular Convection Observed by Magnetic Resonance Imaging, E. E. Ehrichs, H. M. Jaeger, Greg S. Karczmar, James B. Knight, Vadim Yu. Kuperman, and Sidney R. Nagel, Science 267, 16321634 1995. The rst direct observation that convective rolls are present in vertically vibrated granular media, imaged using MRI. E 139. Convection in Horizontally Vibrated Granular Material, Milica Medved, Damien Dawson, Heinrich M. Jaeger, and Sidney R. Nagel, Chaos 9, 691 696 1999. I 140. Transition to Parametric Wave Patterns in a Vertically Oscillated Granular Layer, Francisco Melo, Paul Unbanhowar, and Harry L. Swinney, Phys. Rev. Lett. 72, 172175 1994. Experimental study of surface patterns formed by granular media in wide, shallow containers as a function of vertical oscillation amplitude and frequency. I 141. Localized Excitations in a Vertically Vibrated Granular Layer, P. B. Umbanhowar, F. Melo, and H. L. Swinney, Nature London 382, 793796 1996. The rst experimental report of the observation of spatially localized excitations oscillons in a vertically vibrated shallow granular system. I 142. Spiral Patterns in Oscillated Granular Layers, John R. de Bruyn, B. C. Lewis, M. D. Shattuck, and Harry L. Swinney, Phys. Rev. E 63, 41305 41316 2001. A 143. Electrostatically Driven Granular Media: Phase Transitions and Coarsening, I. S. Aranson, D. Blair, V. A. Kalatsky, G. W. Crabtree, W.-K. Kwok, V. M. Vinokur, and U. Welp, Phys. Rev. Lett. 84, 3306 3309 2000; Phase Separation and Coarsening in Electrostatically Driven Granular Media, I. S. Aranson, B. Meerson, P. V. Sasorov, and V. M. Vinokur, Phys. Rev. Lett. 88, 204301204304 2002. Instead of mechanically vibrating a baseplate, an applied electric eld is employed to oscillate a charged granular medium. A dynamical phase diagram for nucleation and coarsening of dense clusters as a function of frequency and maximum applied electric eld is presented. A 144. Surface Waves in Vertically Vibrated Granular Materials, H. K. Pak and R. P. Behringer, Phys. Rev. Lett. 71, 18321835 1993. Experimental observation of surface waves that propagate up the slope of a sandpile for a granular system in which an annular layer is vertically vibrated. E 145. Segregation Kinetics of Particulate Solids Systems. I. Inuence of Particle Size and Particle Size Distribution, James L. Olsen and Edward G. Rippie, J. Pharm. Sci. 53 2, 147150 1964; Segregation Kinetics of Particulate Solids Systems. II. Particle DensitySize Interactions and Wall Effects, Edward G. Rippie, James L. Olsen, and Morris D. Faiman, ibid. 53 11, 13601363 1964; Segregation Kinetics of Particulate Solids Systems. III. Dependence 18 Am. J. Phys., Vol. 73, No. 1, January 2005

146.

147.

148.

149.

150.

151.

152.

153.

on Agitation Intensity, Morris D. Faiman and Edward G. Rippie, ibid. 54 5, 719722 1965. The rst experimental reports of vertical segregation of granular materials of differing sizes under vertical vibrations, spread out over several publications. Noteworthy is the journal chosen for publication, indicating that this phenomenon is not simply of academic interest but has signicant ramications for industrial powder processing. I Why the Brazil Nuts Are on Top: Size Segregation of Particulate Matter by Shaking, A. Rosato, K. J. Strandburg, F. Prinz, and R. H. Swendsen, Phys. Rev. Lett. 58, 1038 1040 1987. Simulation study of vertical segregation that supported the ratcheting mechanism. As only periodic boundary conditions were investigated, convective rolls would not have been observed. A An Experimental Study of Granular Convection, J. B. Knight, E. E. Ehrichs, V. Yu. Kuperman, J. K. Flint, H. M. Jaeger, and S. R. Nagel, Phys. Rev. E 54, 5726 5738 1996. Thorough overview, using MRI, of the dynamics of convective rolls in vertically vibrated systems, characterizing both the radial and depth dependence of the vertical ow velocity, as the container aspect ratio is systematically varied. A Size Segregation in a Two-Dimensional Sandpile: Convection and Arching Effects, J. Duran, T. Mazozi, E. Clement, and J. Rajchenbach, Phys. Rev. E 50, 5138 5141 1994. Experimental studies of two-dimensional granular media vertically shaken demonstrate that convection and heaping occur for differing regimes of excitation amplitude and frequency. I Density-Noise Power Fluctuations in Vibrated Granular Media, E. R. Nowak, A. Grushin, A. C. B. Barnum, and M. B. Weissman, Phys. Rev. E 63, 2030120304 2001. Experimental studies that nd that the spectral density of density uctuations in vertically vibrated granular systems are characterized by non-Gaussian statistics, indicating the presence of strongly cooperative interactions between uctuators. A Size Segregation and Convection, T. Poschel and H. J. Herrmann, Europhys. Lett. 29, 123128 1995. Molecular-dynamics simulations conrm that convective cells are associated with the rise of larger particles in vertically vibrated systems. A Rise-Time Regimes of a Large Sphere in Vibrated Bulk Solids, L. Vanel, A. D. Rosato, and R. N. Dave, Phys. Rev. Lett. 78, 12551258 1997. Experimental studies of the rise of larger particles embedded in a granular system when vertically vibrated. These studies nd that segregation occurs both for large-amplitude high-frequency oscillations, which induce convective rolls at the sidewalls, as well as for low-amplitude low-frequency excitations that induce a nonconvective regime. A Hydrodynamic Description of Granular Convection, H. Hayakawa, S. Yue, and D. C. Hong, Phys. Rev. Lett. 75, 2328 2331 1995; Trafc Equations and Granular Convection, D. C. Hong and S. Yue, Phys. Rev. E 58, 4763 4775 1998. A hydrodynamic model, along with numerical simulations for granular convection in a vertically vibrated bed is presented. A Size Segregation of Granular Particles, Matthias E. Mobius, Benjamin E. Lauderdale, Sidney R. Nagel, and Heinrich M. Jaeger, Nature London 414, 270 2001. Brief report of experimental study of vertical segregation effects as a function of ambient air pressure in container. The timing and magnitude of the vertical segregation effect are found to decrease at lower background pressures, indicating a role played by the interstitial air. A

VIII. AVALANCHE STRATIFICATION When a mixture of large and small granular media is poured into a vertical HeleShaw cell, consisting of two vertical transparent plates held apart with a narrow separation, mounted on a horizontal baseplate, another striking segregation effect may be observed.154 Typically, these two vertical plates are closed at one end, but this is not crucial to observe the segregation phenomenon. Initially, the granular mixture forms a pile at the bottom of the cell. As the pile grows, two segregation phenomena may be observed, even when care is taken to ensure that the granular mixture is not segregating as it leaves the hopper above the HeleShaw cell. The larger
James Kakalios 18

particles will tend to accumulate at the bottom of the growing pile, leaving the smaller particles separated near the top. Since to rst order the velocity of the falling grains is independent of their size, both large and small grains arrive at the top of the pile with the same velocity. However, their mass difference leads to different kinetic energies. As the particles move in an avalanche down the sandpiles free surface, they suffer inelastic collisions with the other particles in the pile. The larger particles require more collisions to exhaust their initial kinetic energy and are more likely to continue moving down to the bottom of the pile, where they come to rest and accumulate, leaving the smaller particles on top.155 In addition to this segregation, certain granular mixtures can display a stratication effect, whereby the avalanching material separates into alternating layers of large and small grains.154,156 161 There are essentially two mechanisms that have been proposed to account for this stratication effect. One model suggests that stratication does not begin until the pile has reached a size such that even the most energetic particles cannot make it down to the bottom of the sandpile. At this point a metastable wedge forms. When the top surface of this wedge exceeds the maximum angle of stability, the material within the wedge moves in an avalanche down the top surface. As the granular mixture ows, it dilates owing to the shear force arising from the sharp velocity discontinuity between the owing layer and the rigid pile. In this expanded, owing layer, the smaller particles may drop down to the rigid surface beneath the owing layer. The metastable wedge thereby separates into two layers of small particles underneath a layer of large particles.157,160 The alternative model posits that stratication occurs continuously as the material ows down the top surface, with the smaller particles, which are more susceptible to inelastic collisions,162 being caught in small gaps, and stopping rst.156,158,161 The metastable-wedge argument accounts for the observation that stratication only occurs once a critical height of the pile is reached, and that its onset may be varied by changing the kinetic energy of the incoming granular mixture added to the top of the growing pile.160 However, the observation of an upward-moving density wave during the layering process is well described by the continuousow model. Which mechanism dominates turns out to be a sensitive function of the cell geometry and granular-ow rates.160 For certain congurations both mechanisms compete, leading to pairing of small particle bands.157,160 Shear ows as granular material moves in an avalanche down an inclined slope, either in chute ow or along the surface of a sandpile tilted above the angle of maximum stability, can exhibit cluster formation, roll waves, or a ngering instability as the granular front propagates.163
154. The Segregation of Particulate Materials, J. C. Williams, Powder Technol. 15, 245256 1976. The rst experimental report of stratication of binary mixture of granular media when poured into a narrow vertical Hele-Shaw cell. I 155. Interparticle Percolation and Segregation in Granular Materials: A Review, S. B. Savage in Developments in Engineering Mechanics, edited by A. P. S. Selvadurai Elsevier, Amsterdam, 1987, pp. 347 363; Particle Size Segregation in Inclined Chute Flow of Cohesionless Granular Solids, S. B. Savage and C. K. K. Lun, J. Fluid Mech. 189, 311335 1988. Experimental investigation of percolation of ner particles to the bottom of a owing granular layer during an avalanche. I 156. Spontaneous Stratication in Granular Mixtures, Hernan A. Makse, Shlomo Havlin, Peter R. King, and H. Eugene Stanley, Nature London 386, 379382 1997. Experimental report of avalanche stratication and segregation, supported by cellular-automata 19 Am. J. Phys., Vol. 73, No. 1, January 2005

157.

158.

159.

160.

161.

162.

163.

simulations. Proposed mechanism for stratication involves continuous trapping of ner particles during avalanche, with larger grains residing on top. I Phase Diagram for Avalanche Stratication of Granular Media, J. P. Koeppe, M. Enz, and J. Kakalios, Phys. Rev. E 58, R4104 R4107 1998. Experimental investigation and numerical simulations of avalanche stratication as the plate separation of the HeleShaw cell and ow rate of addition of granular mixture are systematically varied. First report of pairing of small particle layers for certain plate separations and ow rates. Model proposed involves the development of metastable wedge between angles of maximum stability and repose. During ow of material in wedge, sieving percolation segregation occurs as described by Savage Ref. 155. I Dynamics of Granular Stratication, H. A. Makse, R. C. Ball, H. Eugene Stanley, and S. Warr, Phys. Rev. E 58, 33573367 1998; Mechanisms of Granular Spontaneous Stratication and Segregation in Two-Dimensional Silos, Pierre Cizeau, Hernan A. Makse, and H. Eugene Stanley, ibid. 59, 4408 4421 1999. Elucidation of cellular-automata model, supported by analytical calculations and experimental observations, in support of continuous-ow model of avalanche stratication. A Stripes Ordering in Self-Stratication Experiments of Binary and Ternary Granular Mixtures, N. Lecocq and N. Vandewalle, Phys. Rev. E 62, 8241 8244 2000. Experimental study of avalanche stratication when ternary granular mixtures are poured into a vertical HeleShaw cell. Depending on the characteristics of the granular material employed, differing layering schema are observed. I Avalanche Stratication-Experimental Tests of the Metastable Wedge and Continuous Flow Models, M. E. Swanson, M. Landreman, J. Michel, and J. Kakalios, Mater. Res. Soc. Symp. Proc. 627, BB2.6 2001. Experimental attempts to reconcile two differing models for avalanche stratication. Which mechanism dominates is sensitive function of plate separations and ow rates. I Microscopic Model for Granular Stratication and Segregation, H. A. Makse and H. J. Herrmann, Europhys. Lett. 43, 1 4 1998. Numerical simulations supporting the continuous-ow model for avalanche stratication. A Different Characteristics of the Motion of a Single Particle on a Bumpy Inclined Line, G. H. Ristow, F.-X. Riguidel, and D. Bideau, J. Phys. I 4, 11611172 1994. Experimental and theoretical investigation of a single ball rolling down a rough inclined plane, to elucidate the energy-dissipation mechanisms during real granular avalanches. A Fingering in Granular Flows, O. Pouliquen, J. Delour, and S. B. Savage, Nature London 386, 816 1997; Segregation Induced Instabilities of Granular Fronts, O. Pouliquen and J. W. Vallance, Chaos 9, 621 630 1999; Longitudinal Vortices in Granular Flows, Y. Forterre and O. Pouliquen, Phys. Rev. Lett. 86, 5886 5889 2001. Experimental studies elucidating mechanisms underlying ngering instability of advancing front of granular material owing down an inclined plane, driven by convection transverse to the direction of downward ow. A

IX. AXIAL SEGREGATION Consider a cylinder partially lled with a binary mixture of granular materials differing in size or density and positioned so that its long axis is parallel to the horizontal plane. The cylinder is now rotated about its long axis, like a drum mixer. If the cylinder is more than 50% lled by volume, then either segregation of the mixture of granular materials about the axis of rotation or if the system is initially prepared in a segregated state, patterning may occur.164 168 This is because only a narrow wedge near the top surface is free to ow as the system is rotated. Geometric shadowing will determine the amount of material affected by the rotation. In this way a central segregated core near the axis of rotation will either become mixed or patterned depending on the extent to which the material within an avalanching wedge intersects the central region. Depending on the conguration of
James Kakalios 19

Fig. 5. Images of a horizontal cylinder, 2 ft long and 5 in. in diameter, one-third lled with a 50/50 mixture by volume of uncooked rice and split peas. The top image is in an initial mixed state and the lower image is of the same system following rotation about the long horizontal axis at 15 rpm for 2 h. See K. M. Hill and J. Kakalios, http://www.physics.umn.edu/groups/ sand Ref. 54.

the container and the amount of material present, this shadowing can lead to a variety of patterns of the binary granular mixtures as the cylinder is rotated. If the system is less than 50% lled with a granular mixture by volume, then two other distinct types of segregation may occur. Initially, within the rst few rotations of the cylinder, a fraction of the smaller particles will concentrate near the axis of rotation.169173 MRI studies have conrmed that this radial segregation occurs along the entire length of the cylinder,174 though initially it had been studied in quasi-twodimensional cylinders that are only a few particle diameters long. This radial segregation results from the sieving of the smaller grains through the dilated, top avalanching surface. In addition, depending on the rotation speed of the cylinder, a second segregation effect may be observed.175186 In this case the granular material forms nearly homogeneous alternating bands of large and small particles, which occur along the length of the cylinder like rings on a nger, as illustrated in Fig. 5. This phenomenon is termed axial segregation. The location and width of the bands varies randomly from trial to trial. MRI studies nd that the radially segregated state persists even in the presence of the axial segregation,174 so that a band of large particles is actually an outer ring of large beads, with a core of small particles near the axis of rotation. Certain systems display a reversible axialsegregation effect, in that a homogeneous mixed state will form alternating segregated bands when the cylinder is rotated at a relatively high rotation speed typically 15 rpm or higher, while if rotation is then continued but at a slower speed such as 5 rpm) the segregated bands disappear and the mixed state is restored.184,185 However, MRI studies nd that even in this re-mixed state the radial segregated core remains.174 The standard model for axial segregation involves the angle the avalanching material makes with the back wall of the cylinder. As the cylinder rotates, the granular material maintains contact with the back wall of the cylinder until the angle of maximum stability is reached, at which point the grains ow down the top surface, and are then brought back
20 Am. J. Phys., Vol. 73, No. 1, January 2005

up again as the cylinder turns. Even at 5 rpm, the granular material is in constant motion down the free surface. Viewed from the side, the granular material forms a dynamic angle of repose, which depends on the properties of the granular material, the rotation speed of the cylinder, the frictional characteristics of the cylinder wall, and the concentration ratio of large to small beads, for example of the mixture. It is found that, for a given rotation speed, a mixture of large and small beads that have the same density mass/bead volume maintains contact with the back wall to a greater height, than for an identical cylinder lled either with all large or all small beads. A collection of all large particles, for example, will have several large voids and arches, while a 50/50 mixture can have these open spaces lled with smaller beads. In this way the granular mixture is more compact and stiffer than the same volume of monodisperse large beads, and hence can be tilted to a larger angle before an avalanche occurs. This concentration dependence of the dynamic angle of repose accounts for the axial-segregation effect. At a relatively high rotation speed, such as 15 rpm, the granular mixture forms a larger dynamic angle of repose than if only large particles are present. However, random collisions between the avalanching beads can result in concentration uctuations, where the number of large beads in one section of the cylinder is greater than the average. In this case the dynamic angle of repose of the section enriched with large particles will be lower than the adjoining regions, which consist of the average mixed state. A large bead at the interface of this uctuation-induced large-bead-enhanced region can therefore lower its gravitational potential energy by falling from the mixed state into this region. In this way the initially narrow large-bead-rich region acquires additional large beads and grows in width. Moreover, as more and more large beads drift from the mixed regions to the large-bead band, the mixed state becomes richer in small beads. In this way, uctuations of more ordered regions that is, concentrations differing from the average will be stabilized and grow, owing to interactions at its interface. Of course, random collisions at the interface also will lead to standard Fickian diffusion, which will tend to re-mix the large-bead band back into the mixed state. When these two effects of gravitational drift and Fickian diffusion are combined, the resulting time dependence for the spatial concentration of large beads is described by a diffusion equation with an effective diffusion coefcient that is the difference of two terms reecting drift and diffusion. At 15 rpm, drift dominates over diffusion and the effective diffusion coefcient is negative, describing a situation where an ordered uctuation is stable and grows in time. For certain granular mixtures at slower rotation speeds such as 5 rpm there is no difference in the dynamic angle of repose between the mixed and all-large-bead states. In this case there is no drift term. If the state is mixed, it will stay mixed, and if axially segregated, random collisions return the alternating bands to a homogeneous mixed state. Systems for which the mixed-state dynamic angle of repose is larger than the large-bead case at all rotation speeds demonstrate a nonreversible axial-segregation effect, and those for which there is no angle difference at any speed never segregate aside from radial segregation.185 While this explanation is consistent with observations of axial segregation and measurements of the concentration dependence of the dynamic angle of repose, MRI studies have found that concentration modulations in the bulk of the granular system may be present that do not extend to the top
James Kakalios 20

of the free-owing rotating surface.174 That is, there can be more axially segregated structure within the bulk than present on the top surface, which casts doubt on a dynamicangle-of-repose model that posits that the segregation mechanism is purely a surface effect, driven by height variations of the top owing layer. Rather, MRI studies indicate that the depth of the avalanching material is greater for binary mixtures than for single-phase regions. This suggests that churning of the radially segregated core by the avalanching wedge induces concentration uctuations, which in turn are reected in the surface dynamic angle of repose.189,190 These height variations may then stabilize and reinforce the concentration modulations, leading to a quasi-stable axially segregated banding pattern. In addition to the radial segregation of smaller particles within the interior of the rotating granular mixture, there is another phenomenon that is observed prior to the formation of stable axially segregated bands. This is transient bands of segregated materials that appear as bidirectional traveling waves when viewed from the top owing surface.191,192 These initial traveling-wave patterns eventually stabilize to form the pattern of axially segregated bands discussed above. Continued rotation for longer times leads to coarsening and evolution of this banding pattern.193,194
164. Avalanche Mixing of Granular Solids, G. Metcalfe, T. Shinbrot, J. J. McCarthy, and J. M. Ottino, Nature London 374, 39 42 1995. Experimental study of geometric shadowing for granular media in a slowly rotating cylinder as a function of lling height. I 165. Size Segregation and Convection of Granular Mixtures Almost Completely Packed in a Thin Rotating Box, Akinori Awazu, Phys. Rev. Lett. 84, 4585 4588 2000. Simulations of a quasi-twodimensional rotating box, where the axis of rotation passes through center of square face of box, nearly full of granular media. A global convection is observed following the formation of the radially segregated state, which in turn induces an axially segregated banding pattern. A 166. Comparing Extremes: Mixing of Fluids, Mixing of Solids, Julio M. Ottino and Troy Shinbrot, in Mixing: Chaos and Turbulence, edited by Chante et al. Kluwer Academic, New York, 1999, pp. 163 186. Comparison of uiduid mixing and granular solidsolid segregation and mixing in rotating horizontal cylinders. Radial and axial segregation are discussed, as the inuence of geometric shadowing of top owing surface intersecting the central stationary core. E 167. Pattern Formation during Mixing and Segregation of Flowing Granular Media, G. Metcalfe and M. Shattuck, Physica A 233, 709 717 1996. MRI study of axial segregation for binary granular mixtures as the particle size and density are systematically varied. A 168. Transverse Flow and Mixing of Granular Materials in a Rotating Cylinder, D. V. Khakhar, J. J. McCarthy, T. Shinbrot, and J. M. Ottino, Phys. Fluids 9, 31 43 1997. Surface ow for monodisperse granular media in a horizontal rotating cylinder is investigated using digital-image analysis. The variation of the average velocity of the owing material as a function of rotation speed, particle properties, and depth toward the axis of rotation is described. I 169. An Analysis of Radial Segregation for Different Sized Spherical Solids in Rotary Cylinders, N. Nityanand, B. Manley, and H. Henein, Metall. Trans. B 17B, 247257 1986. Experimental observations of radial segregation for two-dimensional cylinders. I 170. Particle Mass Segregation in a Two-Dimensional Rotating Drum, G. H. Ristow, Europhys. Lett. 28, 97101 1994. Experimental study of radial segregation in two-dimensional cylinders for binary granular mixtures as the ratio of particle mass is systematically varied. A 171. Radial Segregation in a Two-Dimensional Rotating Drum, C. M. Dury and G. H. Ristow, J. Phys. I 7, 737745 1997. Simulations of radial segregation using the discrete element method. A 172. Radial Segregation in a 2D Drum: Experimental Analysis, F. Cantelaube and D. Bideau, Europhys. Lett. 30, 133138 1995. Experimental study of large and small discs in a two-dimensional i.e., short rotating cylinder. I 21 Am. J. Phys., Vol. 73, No. 1, January 2005

173. Mixing of a Granular Material in a Bi-Dimensional Rotating Drum, E. Clement, J. Rajchenbach, and J. Duran, Europhys. Lett. 30, 712 1995. Using motion capture by a CCD camera, trajectories of tracer particles are measured for rotation in a two-dimensional cylinder. When the tracer is the same size as the other beads, two competing attractive regions, near the cylinder center and the outer wall, are identied. I 174. Bulk Segregation in Rotated Granular Material Measured by Magnetic Resonance Imaging, K. M. Hill, A. Caprihan, and J. Kakalios, Phys. Rev. Lett. 78, 5053 1997. MRI study of concentration variations beneath the top surface for radial and axial segregation in a horizontal cylinder. Study found evidence of segregated banding within the bulk that did not extend to the top surface, raising questions concerning models for axial segregation that suggest the phenomenon is surface-ow-driven. A 175. Horizontal Rotating Cylinder, Y. Oyama, Bull. Inst. Phys. Chem. Res. Jpn. Rep. 18, 600 1939 in Japanese. First experimental report of axial segregation of binary mixture of granular media in a horizontal rotating drum. A 176. Mixing of Solids, S. S. Weidenbaum, Adv. Chem. Eng. 2, 211 1958. I. 177. Mixing and De-Mixing of Solid Particles: Part I. Mechanisms in a Horizontal Drum Mixer, M. B. Donald and B. Roseman, B. Chem. Eng. 7 10 749753 1962; Mixing and De-Mixing of Solid Particles: Part II. Effects of Varying the Operating Conditions of a Horizontal Drum Mixer, 7 11, 823 827 1962. Early experimental study of axial-segregation effect. A model is proposed that is based upon velocity gradients created owing to drag on the granular material by the cylinder walls. I 178. Particle Mixing by Percolation, J. Bridgwater, N. W. Sharpe, and D. C. Stocker, Trans. Inst. Chem. Eng. 47, T114 T119 1969. Experimental studies of mechanisms underlying radial segregation. A 179. Fundamental Powder Mixing Mechanisms, J. Bridgwater, Powder Technol. 15, 215231 1976. I 180. Axial Transport of Granular Solids in Horizontal Rotating Cylinders. Part 1: Theory, S. Das Gupta, D. V. Khakhar, and S. K. Bhatia, Powder Technol. 67, 145151 1991; Axial Segregation of Particles in a Horizontal Rotating Cylinder, Chem. Eng. Sci. 46, 1513 1517 1991; Axial Transport of Granular Solids in Rotating Cylinders. Part 2: Experiments in a Non-Flow System, S. J. Rao, S. K. Bhatia, and D. V. Khakhar, Powder Technol. 67, 153162 1991. Axial segregation effect is described for binary granular mixtures with varying particle diameters. Results are described in terms of a dynamical-angle-of-repose model. I 181. Dynamics of Avalanches in a Rotating Cylinder, S. Fauve, C. Laroche, and S. Douady, in Physics of Granular Media, edited by Daniel Bideau and John Dodds Nova Science, Commack, NY, 1991, p. 277. I 182. Disorder, Diffusion and Structure Formation in Granular Flows, Stuart B. Savage, in Disorder and Granular Media, edited by D. Bideau and A. Hansen North-Holland, Amsterdam, 1993, pp. 255 285. Clear discussion of dynamical-angle-of-repose model for axial segregation, supported by cellular-automata simulations, as well as an exposition of granular ow from a hopper. E 183. Rotationally Induced Segregation of Granular Materials, O. Zik, Dov Levine, S. G. Lipson, S. Shtrikman, and J. Stavans, Phys. Rev. Lett. 73, 644 647 1994. Experimental observation and analytical theory for axial segregation of binary mixtures of granular media in rotating cylinders. A 184. Reversible Axial Segregation of Binary Mixtures of Granular Materials, K. M. Hill and J. Kakalios, Phys. Rev. E 49, R3610R3613 1994. First report of an axial segregation effect that is removed upon continued rotation at slower rotation speeds. Measurements of continuous owing proles support dynamical-angle-of-repose model for segregation effect. E 185. Reversible Axial Segregation of Rotating Granular Media, K. M. Hill and J. Kakalios, Phys. Rev. E 52, 4393 4400 1995. Study of reversible axial segregation effect as the relative diameters of binary mixtures of granular materials are systematically varied. Further experimental support of dynamical-angle-of-repose model for segregation effect is presented. I 186. Axial Segregation of Granular Materials, Dov Levine, Chaos 9, 573580 1999. Review article of advances in understanding of axial segregation. I James Kakalios 21

187. Non-Invasive Measurements of Granular Flows by Magnetic Resonance Imaging, M. Nakagawa, S. A. Altobelli, A. Caprihan, E. Fukushima, and E. K. Jeong, Exp. Fluids 16, 54 60 1993. Early study of sub-surface concentration and velocity variations of granular media in a horizontal rotating cylinder using Magnetic Resonance Imaging. A 188. Axial Segregation of Granular Flows in a Horizontal Rotating Cylinder, Masami Nakagawa, Chem. Eng. Sci. 49, 25402544 1994. MRI study of axial segregation for granular mixture of a continuous size distribution. A 189. Structure and Kinematics in Dense Free-Surface Granular Flow, K. M. Hill, G. Gioia, and V. V. Tota, Phys. Rev. Lett. 91, 64302 64305 2003. Experimental study of the velocity prole of owing beads in a rotating cylinder, demonstrating that the self-diffusion coefcient scales with the beads mean velocity. I 190. Solid-Fluid Transition in a Granular Shear Flow, Ashish V. Orpe and D. V. Khakhar, Phys. Rev. Lett. 93, 68001 68004 2004. Experimental study of velocity prole as a function of depth from the free-surface for granular media in a horizontal rotating cylinder, nding a sharp transition from uid-like behavior to a uid solid state with decreasing velocity beneath the freely owing surface. I 191. Traveling Granular Segregation Patterns in a Long Drum Mixer, Kiam Choo, T. C. A. Molteno, and Stephen W. Morris, Phys. Rev. Lett. 79, 29752978 1997. Experimental observation of travelingwave patterns during initial transients, prior to formation of stable axial segregated band pattern, for long narrow horizontal cylinders. I 192. Axial Segregation of Powders in a Horizontal Rotating Tube, Joel Stavens, J. Stat. Phys. 93, 467 475 1998. A brief review of experimental and theoretical studies of axial segregation. E 193. Axial Segregation of Granular Media in a Drum Mixer: Pattern Evolution, K. M. Hill, A. Caprihan, and J. Kakalios, Phys. Rev. E 56, 4386 4393 1997. Surface observations and MRI studies of evolution of axially segregated banding pattern for extended rotation times. A 194. Avalanche-Mediated Transport in a Rotated Granular Mixture, Vidar Frette and Joel Stavans, Phys. Rev. E 56, 6981 6990 1997. Experimental study of axial-segregation band merging for extended rotation of the horizontal cylinder. The authors propose that axially propagating avalanches are responsible for the band coarsening. I

trafc, have been made by many authors.19,73,185 Indeed, the nonlinear diffusion equation that describes the conditions for which axial segregation is observed has also been employed to describe the spontaneous formation of trafc jams in highway ow.195199 At large auto densities, drivers pack their cars to a point of marginal stability, limited by the drivers awareness of trafc conditions ahead of them and their nite response time. This densely packed state is unstable against uctuations, so that one driver suddenly slowing down or even just tapping his or her brakes can lead to a backward propagating avalanche of stopped cars that the lead car which instigated the jam will not be part of. However, one should approach cautiously this or any other metaphor for granular systems. To say that a sandpile is similar to trafc or that protein dynamics is similar to glassy relaxation is simply transferring ignorance from one eld to another S. R. Nagel, private communication, 1987. Unless the simile leads to new insights into either granular media or highway ow, the mere demonstration that two distinct, complex systems may be described by the same differential equation is not overly instructive. Better to study a sandpile and learn its ways. If nothing else, scientic justications for a trip to the beach are far too rare to be passed up.
195. Trafc Dynamics: Studies in Car Following, R. E. Chandler, R. Herman, and E. W. Montroll, Oper. Res. 6, 165184 1958. Nonlinear diffusion-equation analysis of instabilities in highway ow. I 196. On Kinematic Waves. II. A Theory of Trafc Flow on Long Crowded Roads, M. J. Lighthill and G. B. Whitham, Proc. R. Soc. London, San. A 229, 317345 1955. I 197. Shock Waves on the Highway, Paul I. Richards, Oper. Res. 4, 4251 1956. Analytical modeling of highway ow as a continuous-uid with emphasis on the development of waves in space and time resulting from perturbations such as trafc signals. I 198. Nonlinear Effects in the Dynamics of Car Following, G. F. Newell, Oper. Res. 9, 209229 1961. Analysis of the propagation of smallamplitude disturbances through trafc, when modeled by a nonlinear diffusion equation. I 199. Trafc and Related Self-Driven Many-Particle Systems, D. Helbing, Rev. Mod. Phys. 73, 10671141 2001. A review, for the expert, of the theoretical advancements in understanding trafc dynamics since the pioneering studies in Refs. 195198. A

X. GRANULAR MEDIA AND TRAFFIC The connection between granular media and other complex, nonequilibrium systems, such as glasses or automobile

22

Am. J. Phys., Vol. 73, No. 1, January 2005

James Kakalios

22