Indetermination in Granular Mechanics



Similar documents
Fric-3. force F k and the equation (4.2) may be used. The sense of F k is opposite

APPLIED MATHEMATICS ADVANCED LEVEL

Structural Analysis - II Prof. P. Banerjee Department of Civil Engineering Indian Institute of Technology, Bombay. Lecture - 02

HW Set VI page 1 of 9 PHYSICS 1401 (1) homework solutions

Prelab Exercises: Hooke's Law and the Behavior of Springs

Chapter 11 Equilibrium

C B A T 3 T 2 T What is the magnitude of the force T 1? A) 37.5 N B) 75.0 N C) 113 N D) 157 N E) 192 N

AP Physics C. Oscillations/SHM Review Packet

Physics 9e/Cutnell. correlated to the. College Board AP Physics 1 Course Objectives

Lecture 6. Weight. Tension. Normal Force. Static Friction. Cutnell+Johnson: , second half of section 4.7

NUMERICAL ANALYSIS OF THE EFFECTS OF WIND ON BUILDING STRUCTURES

The Basics of FEA Procedure

Lecture L2 - Degrees of Freedom and Constraints, Rectilinear Motion

AP1 Oscillations. 1. Which of the following statements about a spring-block oscillator in simple harmonic motion about its equilibrium point is false?

Lecture 07: Work and Kinetic Energy. Physics 2210 Fall Semester 2014

E X P E R I M E N T 8

PHY231 Section 2, Form A March 22, Which one of the following statements concerning kinetic energy is true?

PHY231 Section 1, Form B March 22, 2012

Arrangements And Duality

Salem Community College Course Syllabus. Course Title: Physics I. Course Code: PHY 101. Lecture Hours: 2 Laboratory Hours: 4 Credits: 4

Sample Questions for the AP Physics 1 Exam

AP Physics - Chapter 8 Practice Test

Chapter 6. Cuboids. and. vol(conv(p ))

DEM modeling of penetration test in static and dynamic conditions

Chapter 6: The Information Function 129. CHAPTER 7 Test Calibration

Section 1.1. Introduction to R n

Lecture L22-2D Rigid Body Dynamics: Work and Energy

Chapter 6 Work and Energy

COMPETENCY GOAL 1: The learner will develop abilities necessary to do and understand scientific inquiry.

B Answer: neither of these. Mass A is accelerating, so the net force on A must be non-zero Likewise for mass B.

Practice final for Basic Physics spring 2005 answers on the last page Name: Date:

Kinetic Energy (A) stays the same stays the same (B) increases increases (C) stays the same increases (D) increases stays the same.

3600 s 1 h. 24 h 1 day. 1 day

Lab 7: Rotational Motion

Solving Simultaneous Equations and Matrices

Structural Axial, Shear and Bending Moments

Chapter. 4 Mechanism Design and Analysis

Chapter 7: Momentum and Impulse

PHY121 #8 Midterm I

4.2 Free Body Diagrams

Introduction to Engineering System Dynamics

Chapter 4. Forces and Newton s Laws of Motion. continued

Physics 2A, Sec B00: Mechanics -- Winter 2011 Instructor: B. Grinstein Final Exam

A Determination of g, the Acceleration Due to Gravity, from Newton's Laws of Motion

VELOCITY, ACCELERATION, FORCE

Physics 201 Homework 8

Copyright 2011 Casa Software Ltd. Centre of Mass

Numerical Analysis of Independent Wire Strand Core (IWSC) Wire Rope

Identification of Energy Distribution for Crash Deformational Processes of Road Vehicles

Unit 4 Practice Test: Rotational Motion

A vector is a directed line segment used to represent a vector quantity.

521493S Computer Graphics. Exercise 2 & course schedule change

LINES AND PLANES CHRIS JOHNSON

Midterm Solutions. mvr = ω f (I wheel + I bullet ) = ω f 2 MR2 + mr 2 ) ω f = v R. 1 + M 2m

Free Fall: Observing and Analyzing the Free Fall Motion of a Bouncing Ping-Pong Ball and Calculating the Free Fall Acceleration (Teacher s Guide)

USING TRIANGULATIONS IN COMPUTER SIMULATIONS OF GRANULAR MEDIA

NUMERICAL MODELLING OF PIEZOCONE PENETRATION IN CLAY

Simple Harmonic Motion

CHAPTER 5 PREDICTIVE MODELING STUDIES TO DETERMINE THE CONVEYING VELOCITY OF PARTS ON VIBRATORY FEEDER

Review D: Potential Energy and the Conservation of Mechanical Energy

Physics Notes Class 11 CHAPTER 3 MOTION IN A STRAIGHT LINE

Serway_ISM_V1 1 Chapter 4

Vector Algebra II: Scalar and Vector Products

Science Standard Articulated by Grade Level Strand 5: Physical Science

Chapter 10 Rotational Motion. Copyright 2009 Pearson Education, Inc.

TEACHER ANSWER KEY November 12, Phys - Vectors

Statics of Structural Supports

Displacement (x) Velocity (v) Acceleration (a) x = f(t) differentiate v = dx Acceleration Velocity (v) Displacement x

Computer Aided Design (CAD), ME , JHU Professor Dan Stoianovici,

Motion of a Leaky Tank Car

KINEMATICS OF PARTICLES RELATIVE MOTION WITH RESPECT TO TRANSLATING AXES

Precise Modelling of a Gantry Crane System Including Friction, 3D Angular Swing and Hoisting Cable Flexibility

Chapter 07 Test A. Name: Class: Date: Multiple Choice Identify the choice that best completes the statement or answers the question.

Newton s Law of Motion

Minkowski Sum of Polytopes Defined by Their Vertices

Aim : To study how the time period of a simple pendulum changes when its amplitude is changed.

Véronique PERDEREAU ISIR UPMC 6 mars 2013

8.012 Physics I: Classical Mechanics Fall 2008

Lab 8: Ballistic Pendulum

Numerical analysis of boundary conditions to tunnels

State of Stress at Point

At the skate park on the ramp

METHODS FOR ACHIEVEMENT UNIFORM STRESSES DISTRIBUTION UNDER THE FOUNDATION

Acceleration Introduction: Objectives: Methods:

Speed A B C. Time. Chapter 3: Falling Objects and Projectile Motion

Lecture 17. Last time we saw that the rotational analog of Newton s 2nd Law is

PHYS 211 FINAL FALL 2004 Form A

F N A) 330 N 0.31 B) 310 N 0.33 C) 250 N 0.27 D) 290 N 0.30 E) 370 N 0.26

Dynamics. Basilio Bona. DAUIN-Politecnico di Torino. Basilio Bona (DAUIN-Politecnico di Torino) Dynamics / 30

Mechanics lecture 7 Moment of a force, torque, equilibrium of a body

Weight The weight of an object is defined as the gravitational force acting on the object. Unit: Newton (N)

Catapult Engineering Pilot Workshop. LA Tech STEP

Plates and Shells: Theory and Computation - 4D9 - Dr Fehmi Cirak (fc286@) Office: Inglis building mezzanine level (INO 31)

B) 286 m C) 325 m D) 367 m Answer: B

Modeling Mechanical Systems

Back to Elements - Tetrahedra vs. Hexahedra

TWO-DIMENSIONAL FINITE ELEMENT ANALYSIS OF FORCED CONVECTION FLOW AND HEAT TRANSFER IN A LAMINAR CHANNEL FLOW

ELASTIC FORCES and HOOKE S LAW

Angular acceleration α

2.5 Physically-based Animation

Transcription:

Indetermination in Granular Mechanics J. J. Moreau Laboratoire de Mécanique et Génie Civil CNRS-Université Montpellier II, France e-mail: moreau@lmgc.univ-montp2.fr Workshop: Discrete numerical simulations of granular materials Institut Henri Poincaré, Paris, March 16th 18th 2005.

Basic example A rigid rod is inserted with friction between two fixed walls slightly converging upwards. Depending on the intensity of wedging, it may remain at rest or fall down under gravity.

In case of equilibrium under gravity an infinity of values are possible for the contact forces. They don t make an affine manifold in R 4, i.e. the standard discussion of hyperstaticity does not apply.

In case of equilibrium under gravity an infinity of values are possible for the contact forces. They don t make an affine manifold in R 4, i.e. the standard discussion of hyperstaticity does not apply. Instead of linear spaces, Unilateral Mechanics is dominated by convex cones For x and x in C, for λ and λ in ]0, + [ : λx + λ x C (but not necessarily if λ or λ are negative)

The natural measure of the intensity of wedging consists of the values of the contact forces. Knowing these initial values here allows one to predict motion unambiguously.

The natural measure of the intensity of wedging consists of the values of the contact forces. Knowing these initial values here allows one to predict motion unambiguously. In contrast with the 19th century conception of mechanical determinism, according to which initial data should consist only of positions and velocities, we propose to consider contact forces as part of the description of each state.

Using a richer model, viz. taking into account some microscopic longitudinal elasticity of the rod, doesn t essentially change the scheme of information processing as far as only instant t 0 is considered.

Using a richer model, viz. taking into account some microscopic longitudinal elasticity of the rod, doesn t essentially change the scheme of information processing as far as only instant t 0 is considered. In fact, the longitudinal deformation then becomes part of the description of each position. And the given elasticity law connects it with contact forces.

But the microelastic model may also be used to handle hysteretic information. Assume that a recording of the the active forces (gravity and action of an operator) exerted to the rod prior to t 0 is available. By taking into account infinitesimal motions of contacts, this allows one to reconstruct the evolution of contacts forces and finally determine their values at instant t 0.

But the microelastic model may also be used to handle hysteretic information. Assume that a recording of the the active forces (gravity and action of an operator) exerted to the rod prior to t 0 is available. By taking into account infinitesimal motions of contacts, this allows one to reconstruct the evolution of contacts forces and finally determine their values at instant t 0. The possibility of hysteretic analysis does not preclude the existence of variables which, at time t 0, convey a summary of history sufficient for predicting further evolution.

What about computation? Two time-stepping numerical strategies are commonly used in the statics or the dynamics of multicontact systems.

Molecular Dynamics Non-interpenetrability is approximated by steep repulsion forces which enter into action when two bodies come close to each other. Hence the interaction forces (also the frictional ones) are connected with fictitious deviations passed from step to step. Such a numerical scheme complies with the proposition of including these forces in the description of each state.

Contact Dynamics Here the essential nonsmoothness of unilateral constraints and dry friction is faced without any regularizing alteration. The velocity function t u R d is evaluated from step to step through an implicit integration scheme. The position function t q is simply updated at each step.

Contact Dynamics Here the essential nonsmoothness of unilateral constraints and dry friction is faced without any regularizing alteration. The velocity function t u R d is evaluated from step to step through an implicit integration scheme. The position function t q is simply updated at each step. Numerical stability is automatically secured, allowing for much larger time-steps than in MD. But each step of the implicit scheme requires solving a heavy nonsmooth Core Problem, the unknowns of which are the contact forces (or impulsions) and the update of u (zero in case of equilibrium).

Various strategies are available to solve the Core Problem.

Various strategies are available to solve the Core Problem. The most common method consists of an iterative procedure à la Gauss-Seidel. One ranges cyclically through the contacts detected as active. For each of them, one solves a single-contact dynamical problem with all other contact forces treated as known. This yields some updates of the velocity u R d and of the local contact force. Cyclic review is continued until a convergence criterion is fulfilled.

The iterative procedure needs to start from an initial guess of the contact forces. This may consist of zero values, but, when computing the evolution of dense assemblies, one considerably accelerates convergence by taking as initial guess the values calculated for contact forces at the antecedent time-step (for the contacts which were already active).

The iterative procedure needs to start from an initial guess of the contact forces. This may consist of zero values, but, when computing the evolution of dense assemblies, one considerably accelerates convergence by taking as initial guess the values calculated for contact forces at the antecedent time-step (for the contacts which were already active). If this is done, CD computation complies with the recommendation of treating contact forces as part of state description.

Otherwise... the iterative procedure allows one to explore indetermination by surveying the whole set of the solutions to the Core Problem of a selected time-step. 1st technique: Repeat the iterative procedure a large number of times, with initial guesses drawn at random from a plausible set. 2nd technique: From some fixed initial guess, repeat the iterative procedure by changing at random the ordering of the cyclic review of contacts.

Equilibrium of a system of b two-dimensional bodies exhibiting κ points of frictional contact. The collection of contact forces evaluates as a point r in R 2κ

Equilibrium of a system of b two-dimensional bodies exhibiting κ points of frictional contact. The collection of contact forces evaluates as a point r in R 2κ which has to verify: The conditions of static Coulomb friction r C C: a polyhedral cone in R 2κ, the Cartesian product of κ angular regions of the respective copies of R 2. The 3b conditions of equilibrium under the specified loads, defining an affine manifold A with dimension 2κ 3b. The set of the solutions equals the (hyper)polyhedron C A

Collection of circular objects at rest under gravity.

Four other contacts

Two ways of repeating the Gauss-Seidel procedure: On the left: The ordering of contact review is changed at random. On the right: Fixed ordering, but initial guess chosen at random. Same 2D sets, but different ghost views of edges of the 2κ-dimensional polyhedron.

In dynamical evolutions the unknowns of the Core Problem are r R 2κ and the velocity update u R 3b. In the event of a sliding contact, Coulomb s law comes to restrain the contact force more tightly than in static situations. The solution set is not expected to be convex anymore.

Accelerated granular flow on a slope Slope 26 ; grain to grain friction 0.3; ground friction 0.5; restitution 0.

Dispersion of acceleration vectors

Detail of the dispersion of acceleration

Dispersion of contact forces

2D model of a grain pile

Pile is created by pouring grains from a localized source onto some rough horizontal ground. It has been experimentally found that the distribution of ground pressure doesn t relate to the height of material above. A local minimum of pressure may even be found at the vertical of the apex.

This necessarily means that some central portion of the pile has part of its weight supported by arching actions from the surrounding bank. In this 2D simulation, a vertical cut is drawn through the granular matter. The programme calculates the total force transmitted across, i.e. the resultant vector of the contact forces exerted by grains with centers on one side upon antagonists on the other side.

At each time-step, when pile construction is numerically simulated, the message from history consists of two parts: the geometric arrangement of grains the contact forces, as calculated in the antecedent step If the latter are not used to initialize Gauss-Seidel iterations, the calculated transmitted force is affected by the undeterminacy shown above as a cloud.

This example was meant to show that, in situations of this sort, the main part of the message from history is conveyed by the geometric arrangement. The small size of the cloud implies that the indeterminacy arising from ignoring antecedent contact forces doesn t gravely impair the demonstration of arching.

This example was meant to show that, in situations of this sort, the main part of the message from history is conveyed by the geometric arrangement. The small size of the cloud implies that the indeterminacy arising from ignoring antecedent contact forces doesn t gravely impair the demonstration of arching. Actually varying slightly the cut position may produce changes in transmitted force of larger amplitude than the above cloud. In the example only 10 contacts are involved in the force transmission across the cut.

All this refers to the time-stepping approximation. of evolution problems.

All this refers to the time-stepping approximation. of evolution problems. Coming to the theoretical formulation of these problems, it doesn t make sense to distinguish input and output values of contact forces. What the preceding reflects is an incremental version of the law of dry friction.

What if friction is zero? A conjecture (J.-N. Roux, C. Moukarzel): Generically, a collection of N frictionless round objects in equilibrium is isostatic.

What if friction is zero? A conjecture (J.-N. Roux, C. Moukarzel): Generically, a collection of N frictionless round objects in equilibrium is isostatic. In 2D models, a frictionless disk has two effective degrees of freedom. Hence a collection of N disks with κ contacts obeys 2N independent equilibrium equations involving κ unknown normal contact forces. Isostaticity requires κ = 2N

A misleading example: Hierarchical deposition, a geometric procedure for constructing sequentially, in 2D, an assembly of contacting disks. Each added disk touches two precedingly deposited ones, hence κ = 2N. Values of frictionless contact forces securing the equilibrium of the whole under given loads are uniquely determined through sequential calculation. But they have not necessarily the directions allowed by unilaterality.

hierachical deposit (31 disks, 62 contacts) frictionless equilibrium under gravity (same)

In view of translational invariance, the effective number of DOF equals 67

As a rigid body in space, the table possesses 6 DOF; its position relative to ground is invariant under displacements parallel to this plane. Hence the effective number of DOF equals 3. The above case of the table having exactly 4 contacts with ground is exceptional.

Proposed explanation:

However there may exist loose grains which make equality κ = 2N not valid anymore. This is related to some of the functions f i being non-strictly convex. Should the exact orthogonality of gravity to ground be considered as exceptional?

Indetermination due to dry friction Halsey, T. C. & Erta, D., A ball in a groove, Phys. Rev. Letters, 83, 5007 5010, 1999. Moreau, J. J., Indétermination liée au frottement sec dans le calcul des granulats, in (M. Potier-Ferry, M. Bonnet, A. Bignonnet, coordonnateurs), Actes du 6ème Colloque National en Calcul des Structures, 2003, vol. 3, pp. 465 472. Unger, T., Kertész, J. & Wolf, D.E., Force indeterminacy in the jammed state of hard disks, ArXiv:cond-mat/0403089, 2004.

Moreau J. J., Indetermination due to dry friction in multibody dynamics. In (P. Neittaanmäki et al., eds.) European Congress on Computational Methods in Applied Sciences and Engineering (ECCOMAS 2004), Jyväskylä, Finland, 24 28 July 2004, vol. 2, CD ROM, ISBN 951-39-1869-6.

On the CD method (and the example of grain piles) J. J. Moreau, Modélisation et simulation de matériaux granulaires, in: (B. Mohammadi, ed.) Actes du 35e Congrès National d Analyse Numérique, Montpellier, 2-6 juin 2003, 30 pages (CD ROM)., An introduction to unilateral dynamics, in: (M. Frémond & F. Maceri, eds.) Novel approaches in Civil Engineering, Lecture Notes in Applied and Computational Mechanics, vol. 14, Springer-Verlag, 2004, pp. 1 46.

Isostaticity in frictionless assemblies Moukarzel, C., Response functions in isostatic packings, in (H. Hinrichsen, D. Wolf eds)the physics of granular media, Wiley-Vch, Berlin, 2005 Roux, J.-N., Geometric origin of mechanical properties of granular materials, Phys. Rev. E, 61 (6), 6802 6836, 2000.