The Finite Element Method

Similar documents
Numerical Heat and Mass Transfer

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

Finite Element Modelling of truss/cable structures

Professor Terje Haukaas University of British Columbia, Vancouver The Q4 Element

Lecture 12: Discrete Laplacian

Module 3: Element Properties Lecture 1: Natural Coordinates

NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS

A boundary element method with analytical integration for deformation of inhomogeneous elastic materials

DUE: WEDS FEB 21ST 2018

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION

Numerical Transient Heat Conduction Experiment

High resolution entropy stable scheme for shallow water equations

Integrals and Invariants of Euler-Lagrange Equations

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

In this section is given an overview of the common elasticity models.

STUDY ON TWO PHASE FLOW IN MICRO CHANNEL BASED ON EXPERI- MENTS AND NUMERICAL EXAMINATIONS

Linear Approximation with Regularization and Moving Least Squares

2.29 Numerical Fluid Mechanics Fall 2011 Lecture 12

Chapter 11: Simple Linear Regression and Correlation

Lecture 13 APPROXIMATION OF SECOMD ORDER DERIVATIVES

FUZZY FINITE ELEMENT METHOD

Report on Image warping

MATH 5630: Discrete Time-Space Model Hung Phan, UMass Lowell March 1, 2018

A Cartesian-grid integrated-rbf method for viscoelastic flows

FTCS Solution to the Heat Equation

Appendix B. The Finite Difference Scheme

Application of B-Spline to Numerical Solution of a System of Singularly Perturbed Problems

Lecture 5.8 Flux Vector Splitting

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

Physics 5153 Classical Mechanics. D Alembert s Principle and The Lagrangian-1

Principles of Food and Bioprocess Engineering (FS 231) Solutions to Example Problems on Heat Transfer

The Two-scale Finite Element Errors Analysis for One Class of Thermoelastic Problem in Periodic Composites

APPENDIX A Some Linear Algebra

New Method for Solving Poisson Equation. on Irregular Domains

NUMERICAL DIFFERENTIATION

Analytical Gradient Evaluation of Cost Functions in. General Field Solvers: A Novel Approach for. Optimization of Microwave Structures

Higher Order Wall Boundary Conditions for Incompressible Flow Simulations

COMPOSITE BEAM WITH WEAK SHEAR CONNECTION SUBJECTED TO THERMAL LOAD

CHAPTER 6. LAGRANGE S EQUATIONS (Analytical Mechanics)

Note 10. Modeling and Simulation of Dynamic Systems

COMPUTATIONAL METHODS AND ALGORITHMS Vol. II - Finite Element Method - Jacques-Hervé SAIAC

The Finite Element Method: A Short Introduction

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Physics 207: Lecture 20. Today s Agenda Homework for Monday

Integrals and Invariants of

Modeling of Dynamic Systems

Introduction to Computational Fluid Dynamics

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Physics 181. Particle Systems

Frame element resists external loads or disturbances by developing internal axial forces, shear forces, and bending moments.

Chapter 4 The Wave Equation

ONE DIMENSIONAL TRIANGULAR FIN EXPERIMENT. Technical Advisor: Dr. D.C. Look, Jr. Version: 11/03/00

DETERMINATION OF TEMPERATURE DISTRIBUTION FOR ANNULAR FINS WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY BY HPM

ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM

ME 501A Seminar in Engineering Analysis Page 1

Curve Fitting with the Least Square Method

Optimal Control of Temperature in Fluid Flow

Formal solvers of the RT equation

Chapter 12. Ordinary Differential Equation Boundary Value (BV) Problems

Construction of Serendipity Shape Functions by Geometrical Probability

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

arxiv: v1 [physics.flu-dyn] 16 Sep 2013

2 Finite difference basics

Thermal-Fluids I. Chapter 18 Transient heat conduction. Dr. Primal Fernando Ph: (850)

Module 1 : The equation of continuity. Lecture 1: Equation of Continuity

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

Appendix for Causal Interaction in Factorial Experiments: Application to Conjoint Analysis

Aerodynamics. Finite Wings Lifting line theory Glauert s method

Computational Electromagnetics in Antenna Analysis and Design

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

STATIC ANALYSIS OF TWO-LAYERED PIEZOELECTRIC BEAMS WITH IMPERFECT SHEAR CONNECTION

χ x B E (c) Figure 2.1.1: (a) a material particle in a body, (b) a place in space, (c) a configuration of the body

Errors for Linear Systems

A Hybrid Variational Iteration Method for Blasius Equation

Chapter Eight. Review and Summary. Two methods in solid mechanics ---- vectorial methods and energy methods or variational methods

Difference Equations

Comparison of Regression Lines

CME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 13

A PROCEDURE FOR SIMULATING THE NONLINEAR CONDUCTION HEAT TRANSFER IN A BODY WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY.

Inductance Calculation for Conductors of Arbitrary Shape

APPENDIX F A DISPLACEMENT-BASED BEAM ELEMENT WITH SHEAR DEFORMATIONS. Never use a Cubic Function Approximation for a Non-Prismatic Beam

Application of Finite Element Method (FEM) Instruction to Graduate Courses in Biological and Agricultural Engineering

Global Sensitivity. Tuesday 20 th February, 2018

1 Convex Optimization

(Online First)A Lattice Boltzmann Scheme for Diffusion Equation in Spherical Coordinate

ALGORITHM FOR THE CALCULATION OF THE TWO VARIABLES CUBIC SPLINE FUNCTION

Chapter 9: Statistical Inference and the Relationship between Two Variables

A comprehensive study: Boundary conditions for representative volume elements (RVE) of composites

Visco-Rubber Elastic Model for Pressure Sensitive Adhesive

Numerical Solutions of a Generalized Nth Order Boundary Value Problems Using Power Series Approximation Method

Interconnect Modeling

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

Notes on Analytical Dynamics

The Discretization Process

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

Robust Norm Equivalencies and Preconditioning

PART 8. Partial Differential Equations PDEs

Kernel Methods and SVMs Extension

Normally, in one phase reservoir simulation we would deal with one of the following fluid systems:

Transcription:

The Fnte Element Method GENERAL INTRODUCTION Read: Chapters 1 and 2 CONTENTS Engneerng and analyss Smulaton of a physcal process Examples mathematcal model development Approxmate solutons and methods of approxmaton The basc features of the fnte element method Examples Fnte element dscretzaton Termnology Steps nvolved n the fnte element model development

INTRODUCTORY REMARKS What we do as engneers? develop mathematcal models, conduct physcal experments, carry out numercal smulatons to help desgner, and desgn and buld systems to acheve a (1) functonalty n (2) most economcal way. Knowng the fundamentals assocated wth each engneerng problem you set out to tackle, not only makes you a better engneer but also empowers you as an engneer. Introducton: 2

INTRODUCTORY REMARKS Engneerng s the dscplne, art, and professon of acqurng and applyng techncal, scentfc, and mathematcal knowledge to desgn and mplement materals, structures, machnes, devces, systems, and processes that safely realze a desred objectve. Engneerng s a problem-solvng dscplne, and soluton requres an understandng of the phenomena that occurs n the system or process. Introducton: 3

Analyss Analyss s an ad to desgn and manufacturng, and not an end n tself. Analyss nvolves the followng steps: dentfyng the problem and nature of the response to be determned, selectng the mathematcal model (.e., governng equatons), solvng the problem wth a soluton method (e.g., FEM), and evaluatng the results n lght of the desgn parameters. Introducton: 4

MODELING OF A Physcal Process Assumptons concernng the system BVP Boundary value problems (equlbrum problems) IVP Intal value problems (tme-dependent problems) Numercal method (FEM,FDM,BEM,etc.) Physcal System Mathematcal Model (BVP, IVP) Numercal Smulatons Laws of physcs (conservaton prncples) FEM Fnte Element Method FDM Fnte Dfference Method BEM Boundary Element Method Computatonal devce Introducton: 5

EXAMPLES OF MATHEMATICAL MODEL DEVELOPMENT Objectve: Determne heat flow n a heat exchanger fn Lateral surface and rght end are exposed to ambent temperature, T (a) Rectangular fns Body from whch heat has to be extracted a h L (b) 3D to 2D y z x Convecton, ( qa ) ( ) x qa x + Δ x ( -T ), P = permeter β P T a æax + A ö x+ Δx ( qa) x -( qa) x+ Δx -βpδx ( T - T ) + ρr hç Δx = 0 çè 2 ø l 2D to 1D d dt - ( Aq) -βp( T - T ) + ρrh A = 0, q =-k dx dx Basc Concepts: 6

EXAMPLE OF ENGINEERING MODEL DEVELOPMENT (contnued) f, force per unt length d æ dtö - ç Ak + βp( T - T ) = ρrh A dx çè dx ø u= T - T, a= Ak, c= βp, ρrh A = f d æ duö - ç a + cu= f dx çè dx ø Determne: Axal deformaton of a bar x Δx L ( Aσ ) x d ( Aσ) x+ Δx -( Aσ) x + faδx = 0 ( Aσ) + fa= 0, dx du d æ duö σ = Eε, ε =, - ç AE + f = 0 dx dx çè dx ø Δx f ( Aσ ) x + Δ x Basc Concepts: 7

Model Problem a = E, A APPROXIMATE SOLUTIONS d du a( x) + c( x) u f ( x) = 0 n Ω = ( 0, L) dx dx du a + b( u u0 ) = P at a boundary pont dx EA f( x) u0 P u(l) unnsulated bar a = ka c= P f( x) β L x, u c b=k Elastc deformaton of a bar x L u= T - T, u = T 0 Heat transfer n a fn Basc Concepts: 8

ENGINEERING EXAMPLES OF THE MODEL PROBLEM IN 1-D Flow of vscous flud through a channel d dv x μ f 0 n ( 0, b) dy = Ω = dy uy ( ) = vx ( y), horzontal velocty f = pressure gradent, dp/ dx(constant) μ( y) = flud vscosty Q = shear stress U = velocty of the top surface y b 2 U ³ 0 b x Poseulle flow Couette flow Basc Concepts: 9

Exact and Approxmate Solutons An exact soluton satsfes (a) the dfferental equaton at every pont of the doman and (b) boundary condtons on the boundary. An approxmate soluton satsfes the dfferental equaton as well as the specfed boundary condtons n some acceptable sense (to be made clearer shortly). We seek the approxmate soluton as a lnear combnaton of unknown parameters c and known functons f ux ( )» u ( x) = å cf ( x) + f ( x) Approxmate soluton and f 0 N N = 1 0 Actual soluton Approxmaton of the actual soluton over the entre doman Basc Concepts: 10

Determnng Approxmate Solutons (contnued) f 1. Suppose that s selected to satsfy the boundary condtons exactly. Then substtuton of u N (x) nto the dfferental equaton d æ du ax N ö - ç ( ) + cxu ( ) N - fx ( ) = 0 dx çè dx ø wll result n a non-zero functon on the left sde of the equalty: d æ du a( x ) N ö - ç + c( x )u N - f( x ) º R( x ) ¹ dx çè dx ø Then c are determned such that the resdual, R(x), s zero n some sense. 0 Basc Concepts: 11

METHODS OF APPROXIMATION 1. One sense n whch the resdual, R(x), can be made zero s to requre t to be zero at selected number of ponts. The number of ponts should be equal to the number of unknowns n the approxmate soluton N ux ( ) u = c f ( x) + f0( x) N j j j = 1 fj ( x) and f0( x) are functons to be selected to satsfy the specfed boundary condtons and cj are parameters to be determned such that the resdual s made zero n some sense. Ths way of determnng c s known as the Collocaton method. We obtan N algebrac equatons n N unknown C s R( x )= 0, = 1, 2,, N Basc Concepts: 12

Methods of Approxmaton (Contnued) 2. Another approach n whch the resdual, R(x), can be made zero s n a least-squares sense;.e., mnmze the ntegral of the square of the resdual wth respect to C s. Mnmze or J ( c J c 1, c = 2 2,, c 0 ) R R c Ths method s known as the least-squares method. We obtan N algebrac equatons n N unknown C s L N = dx L 0 = R 0 2 dx L 0 R R c dx = 0 Basc Concepts: 13

Methods of Approxmaton (Contnued) 3. Yet, another approach n whch the resdual, R(x), can be made zero s n a weghted-resdual sense L 0 = y Rdx, = 1,2,, N ò 0 where y are lnearly ndependent set of functons Ths method s known as the Weghted-Resdual method. We obtan N algebrac equatons n N unknown C s. In general, weght functons ψ are not the same as the approxmaton functons. Varous specal cases are ϕ Petrov-Galerkn Method: Galerkn Method: y ¹f Basc Concepts: 14 y =f

WEIGHTED-INTEGRAL FORMULATIONS n the Numercal Soluton of Dfferental Eqs. The approxmaton methods dscussed earler can be vewed as specal cases of the weghted-resdual methods of approxmaton. In partcular, we have Collocaton method Least-squares method Petrov-Galerkn method Galerkn Method y() x = d( x-x ) R y ( x) = c y ( x) ¹ f ( x) y ( x) = f ( x) Basc Concepts: 15

Methods of Approxmaton (Contnued) 4. Another approach n whch the governng equaton s cast n a weak-form and the weght functon s taken the same as the approxmaton functon s known as the Rtz method: B( f, u ) = ( f ), = 1,2,, N N The Rtz method s the most commonly used method for all commercal software. In ths method, f 0 satsfes only the specfed essental (geometrc) boundary condtons whle f satsfes the homogeneous form of the specfed essental boundary condtons. The specfed natural (force) boundary condton are ncluded n the weak form. Introducton: 16

WORKING EXAMPLE For a weghted-resdual method: f 0 f satsfy the actual specfed BCs satsfy the homogeneous form of the actual specfed BCs For the Rtz method: f 0 satsfy the actual specfed essental BCs f satsfy the homogeneous form of the actual specfed essental BCs d æ duö - ç a - f0 = 0, 0< x< L dx çè dx ø du u(0) = u0, a = P dx x = L Introducton: 17

f WORKING EXAMPLE (CONTINUED) For a weghted-resdual method: df P ( 0) = u, a = P f = u + x a 0 0 0 0 0 dx x= L df f ( 0) = 0, a = 0 f1 = x( 2L-x) dx x= L For the Rtz (or weak-form Galerkn) method: f ( 0) = u, f( 0) = 0 0 0, ( ) f0 = u0 f x = x Introducton: 18

BASIC FEATURES OF THE FINITE ELEMENT METHOD (FEM) Dvde whole nto parts (fnte element mesh) Set up the `problem over a typcal part (derve a set of relatonshps between prmary and secondary varables) Assemble the parts to obtan the soluton to the whole Basc Concepts: 19

Example 1: Determne the center of mass of a 3D machne part 1. See the smplcty n the complcated (see the geometrc shapes that are smple to dentfy the center of mass). 2. Determne the center of mass of each part. 3. Put the parts together to obtan the requred soluton. N N N mx my mz X = Y = Z = m m m = 1 = 1 = 1,, N N N = 1 = 1 = 1 Basc Concepts: 20

Example 2: Determne the ntegral of a functon b I = F( x) a dx F(x) Actual functon a b x Basc Concepts: 21

EXAMPLE 2 (contnued) F( x) a1 a2 a + b x, 3 1 + b 2 + b 3 x, x, a x x x x x 2 x 3 2 3 x b x+ 1 x+ 1 I = F ( x) dx = ( a + b x) dx x x I I + 1 + I 2 I3 F(x) Actual functon Approxmaton I1 I I 2 3 a = x b = x4 x 1 x2 3 x Basc Concepts: 22

EXAMPLE 2 (contnued) Refned mesh F(x) Actual functon Approxmaton a = x 1 7 b = x x Basc Concepts: 23

Approxmaton of a curved surface wth a plane (temperature profle) (represents the soluton) Doman (Trangular element) Basc Concepts: 24

Fnte Element Dscretzaton Doman, Ω Elements Boundary, Γ e Ω Nodes (a) Gven doman (b) Fnte element mesh Boundary flux Ω e Doman, Ω h (c) Typcal element wth boundary fluxes (d) Dscretzed doman Basc Concepts: 25

Some Examples of Real-World Fnte Element Dscretzatons Introducton: 26

FEM Termnology Element A geometrc sub-doman of the regon beng smulated, wth the property that t allows a unque (1) representaton of ts geometry and (2) dervaton of the approxmaton (nterpolaton) functons. Node A geometrc locaton n the element whch plays a role n the dervaton of the nterpolaton functons and t s the pont at whch soluton s sought. Mesh A collecton of elements (or nodes) that replaces the actual doman. Weak Form An ntegral statement equvalent to the governng equatons and natural boundary condtons. More to come. Basc Concepts: 27

FEM Termnology (contnued) Fnte Element Model A set of algebrac equatons relatng the nodal values of the prmary varables (e.g., dsplacements) to the nodal values of the secondary varables (e.g., forces) n an element. Fnte element model s NOT the same as the fnte element method. There s only one fnte element method but there can be more than one fnte element model of a problem (dependng on the approxmate method used to derve the algebrac equatons). Numercal Smulaton Evaluaton of the mathematcal model (.e., soluton of the governng equatons) usng a numercal method and computer. Basc Concepts: 28

Major Steps of Fnte Element Model Development Begn wth the governng equatons of the problem Develop ts weak form over a typcal fnte element Approxmate the soluton over each fnte element Obtan algebrac relatons among the quanttes of nterest over each fnte element (.e., fnte element model) Basc Concepts: 29

Major Steps of Fnte Element Model Development Engneerng Problem Formulaton Governng (Dfferental) Equatons Sold mechancs Weak Form Development Vrtual work statements Fnte Element Model Development Basc Concepts: 30