Lecture 2. Introduction to FEM. What it is? What we are solving? Potential formulation Why? Boundary conditions

Similar documents
Chapter 1 Mathematical Foundations

Antennas and Propagation. Chapter 2: Basic Electromagnetic Analysis

TECHNO INDIA BATANAGAR

ELE3310: Basic ElectroMagnetic Theory

1 Fundamentals. 1.1 Overview. 1.2 Units: Physics 704 Spring 2018

+ f f n x n. + (x)

Chapter 5. Magnetostatics

Chapter 7. Time-Varying Fields and Maxwell s Equations

Mathematical Notes for E&M Gradient, Divergence, and Curl

Transmission Lines and E. M. Waves Prof. R. K. Shevgaonkar Department of Electrical Engineering Indian Institute of Technology, Bombay

Radiation Integrals and Auxiliary Potential Functions

4 Finite Element Analysis of a three-phase PM synchronous machine

Chapter 7. Time-Varying Fields and Maxwell s Equation

INTRODUCTION to the DESIGN and FABRICATION of IRON- DOMINATED ACCELERATOR MAGNETS

r r 1 r r 1 2 = q 1 p = qd and it points from the negative charge to the positive charge.

Electromagnetic Waves For fast-varying phenomena, the displacement current cannot be neglected, and the full set of Maxwell s equations must be used

Electromagnetic Theory Prof. D. K. Ghosh Department of Physics Indian Institute of Technology, Bombay

Unit-1 Electrostatics-1

Electromagnetic Theory Prof. D. K. Ghosh Department of Physics Indian Institute of Technology, Bombay

CHAPTER 2. COULOMB S LAW AND ELECTRONIC FIELD INTENSITY. 2.3 Field Due to a Continuous Volume Charge Distribution

Chapter 5. Magnetostatics

KINGS COLLEGE OF ENGINEERING DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING QUESTION BANK

Time-Varying Systems; Maxwell s Equations

DIVERGENCE AND CURL THEOREMS

Theory of Electromagnetic Nondestructive Evaluation

The Magnetic Field

UNIT I ELECTROSTATIC FIELDS

Homogenization of the Eddy Current Problem in 2D

A Brief Revision of Vector Calculus and Maxwell s Equations

Lecture 10: Vector Calculus II

Motor-CAD combined electromagnetic and thermal model (January 2015)

Gauss Law. In this chapter, we return to the problem of finding the electric field for various distributions of charge.

Elements of Vector Calculus : Scalar Field & its Gradient

Take-Home Exam 1: pick up on Thursday, June 8, return Monday,

MATH 308 COURSE SUMMARY

Electromagnetic wave propagation. ELEC 041-Modeling and design of electromagnetic systems

Introduction to Electromagnetic Theory

Scientific Computing

EECS 117 Lecture 19: Faraday s Law and Maxwell s Eq.

UNIT-I INTRODUCTION. 1. State the principle of electromechanical energy conversion.

fiziks Institute for NET/JRF, GATE, IIT-JAM, JEST, TIFR and GRE in PHYSICAL SCIENCES

General review: - a) Dot Product

MATH 2400: Calculus III, Fall 2013 FINAL EXAM

Principles of Mobile Communications

Electromagnetic Forces on Parallel Current-

Antenna Theory (Engineering 9816) Course Notes. Winter 2016

PHY752, Fall 2016, Assigned Problems

Multiple Integrals and Vector Calculus (Oxford Physics) Synopsis and Problem Sets; Hilary 2015

Mixed Finite Elements Method

ELECTRO MAGNETIC FIELDS

Vector Calculus, Maths II

Physics 6303 Lecture 2 August 22, 2018

Field and Wave Electromagnetic

Differential Operators and the Divergence Theorem

Notes 19 Gradient and Laplacian

Weighted Regularization of Maxwell Equations Computations in Curvilinear Polygons

Brief Review of Vector Algebra

Applied'&'Computa/onal'Electromagne/cs (ACE) Part/III Introduc8on/to/the/Finite/Element/Technique/for/ Electromagne8c/Modelling

43.1 Vector Fields and their properties

Multiple Integrals and Vector Calculus: Synopsis

Determination of a Synchronous Generator Characteristics via Finite Element Analysis

CONSIDER a simply connected magnetic body of permeability

8.03 Lecture 12. Systems we have learned: Wave equation: (1) String with constant tension and mass per unit length ρ L T v p = ρ L

INTRODUCTION ELECTROSTATIC POTENTIAL ENERGY. Introduction. Electrostatic potential energy. Electric potential. for a system of point charges

Electromagnetic Fields. Lecture 2. Fundamental Laws

AN EXAMPLE OF SYSTEM WHICH CAN BE USED TO EXPLICITLY SHOW THE SYMMETRY BETWEEN THE ELECTRIC AND MAGNETIC FIELDS. Arief Hermanto

ELECTROMAGNETIC FIELD

(Refer Slide Time: 03: 09)

Chap. 1 Fundamental Concepts

Foundations of Geomagnetism

UNIT-III Maxwell's equations (Time varying fields)

Electromagnetic Field Theory Chapter 9: Time-varying EM Fields

Overview in Images. S. Lin et al, Nature, vol. 394, p , (1998) T.Thio et al., Optics Letters 26, (2001).

Finite Element Modeling of Electromagnetic Systems

Engineering Electromagnetic Fields and Waves

Electromagnetic Theory Prof. D. K. Ghosh Department of Physics Indian Institute of Technology, Bombay

2577. The analytical solution of 2D electromagnetic wave equation for eddy currents in the cylindrical solid rotor structures

Lecture 13: Electromagnetic Theory Professor D. K. Ghosh, Physics Department, I.I.T., Bombay. Poisson s and Laplace s Equations

09 The Wave Equation in 3 Dimensions

xkcd.com It IS about physics. It ALL is.

Math Divergence and Curl

February 27, 2019 LECTURE 10: VECTOR FIELDS.

free space (vacuum) permittivity [ F/m]

Electromagnetic waves in free space

MATH Max-min Theory Fall 2016

UNIT-I INTRODUCTION TO COORDINATE SYSTEMS AND VECTOR ALGEBRA

Module 3: Electromagnetism

Nitsche-type Mortaring for Maxwell s Equations

Haus, Hermann A., and James R. Melcher. Electromagnetic Fields and Energy. Englewood Cliffs, NJ: Prentice-Hall, ISBN:

Magnetic circuits - electromagnet

Antennas and Propagation

Magnetostatic fields! steady magnetic fields produced by steady (DC) currents or stationary magnetic materials.

Magnetostatic Fields. Dr. Talal Skaik Islamic University of Gaza Palestine

DHANALAKSHMI SRINIVASAN INSTITUTE OF RESEARCH AND TECHNOLOGY

Simulation and Visualization of Safing Sensor

Maxwell s Equations. In the previous chapters we saw the four fundamental equations governging electrostatics and magnetostatics. They are.

Limits to Statics and Quasistatics

Analytical and numerical computation of the no-load magnetic field in induction motors

HIGH VOLTAGE TECHNIQUES REVİEW: Electrostatics & Magnetostatics

3 rd ILSF Advanced School on Synchrotron Radiation and Its Applications

Transcription:

Introduction to FEM What it is? What we are solving? Potential formulation Why? Boundary conditions Lecture 2

Notation Typical notation on the course: Bolded quantities = matrices (A) and vectors (a) Unit vector = e z x = position = (x,y) Operators (click for explanatory material) Gradient of scalar function f = x fe x + y fe y + z fe z Divergence of vector function f = f x e x + f y e y + f y e y f = x f x + y f y + z f z Curl of vector function f = e x e y e z x y z f x f y f z

Notation Special cases: Curl of z-directional vector f = fe z f = f y e x f x e y Curl of a 2D vector (xy-plane) f(x, y) = f x e x + f y e y f = f y x f x y e z Matrix transpose A T and inverse A 1 Dot product of two vectors a b Good to know: Curl of any gradient is always zero: φ 0 Divergence of any curl is always zero: w 0 Wikipedia: Vector calculus identities

Idea of FEM We want to find a function A(x) that satisfies some partial differential equation on some domain (Ω) + some boundary conditions For (a convenient) example ν x A x = J x Known functions of the position x

Idea of FEM ν x A x = J x IDEA: write A(x) as a weighted sum of some known functions N(x) and some unknown coefficients a: n A x a j N j x = A(x) j=1 The functions N are typically called shape functions (or trial functions) Compare to Fourier series The problem of finding an unknown function now reduced to finding unknown scalar coefficients

Idea of FEM In other words: we want to find the coefficients a n A x = a j N j x j=1 in such a way that the equation ν A J is satisfied as well as possible Please note that the end result still is a function It s simply defined by a set of scalar coefficients Takeaway = we have an approximation for A(x) at each and any x useful later

What we are solving Next: where does the div-grad equation come from? ν A = J

What we are solving? Five field variables (E, D, H, B, J) are needed to present a complete electromagnetic field. D B 0 B E t D H J t Maxwell s equations D E J E B H Material equations + Boundary conditions are needed

Boundary conditions = some info about the solution on the outer boundary E.q. values, derivatives, some combination Have to know what happens here Why we need these? Because of (rather complex) maths = if we leave part of the boundary free, we get infinitely many solutions satistying the fixed part + Maxwell s eqs inside. Maxwell s equations satisfied in here

Interface conditions for magnetics Boundaries inside problem domain 1 The tangential component of field strength is continuous H H t1 H t2 Comes from H = 0 (on the thin boundary) B The normal component of flux density is continuous B = μh B n1 B n2 Comes from B = 0 Takeaway: the normal component of H and tangential component of B often have jumps whenever μ does.

Reluctivity In electromechanics, the reluctivity function ν is often used instead of the permeability H = νb ν = 1 μ Easier notation

Simplifications related to timedependence Dynamic problems: Complete Maxwell s equations needed (wave equation) Quasi-static problems: Displacement current neglected D 0 t Static problems: Induced electric field is also neglected B 0 t

Next Introducing the vector potential Definition Properties

Vector potential Problem: Even in static case, we have two equations for one unknown υb = J B = 0 Ampere s law Gauss s law Let s define B by a vector potential A(x) B = A Now Gauss s law is satisfied automatically since any vector A satisfies ( cuz maths) A = 0

Vector potential By definition, the flux density B is B Bxex Byey A A x, y ez A Bx By y A x

Vector potential Flux lines In 2D, the vector potential is constant in the direction of the flux density B This can be seen by computing the directed derivative (how much A changes in direction of B): A A A B = A A = x A y y A x = 0

Vector potential Flux lines In other words: the flux lines correspond to the equipotential lines of A = isolines = contour lines = lines at which A is constant Compare to the contours ( height lines ) in a map Values of A Flux lines

Vector potential Note: vector potential is not unique in 3D * * Several A give the same B A Af => A Af Af A A gauge condition is often included to ensure uniqueness A = 0 In 2D, it is satisfied automatically no need for concern

Vector potential formulation Let s combine the potential with Maxwell s equations: Partial differential equation for the vector potential H J H B => B J => A J B A Nice to know: called curl-curl equation

Formulation in 2D Current density and vector potential point in the same direction J J x, y ez A Ax, ye z Unit vector of z-direction

By writing open Formulation in 2D υ A e z = Je z we get or simply A A e x z Je x y y z ( A) J There s our original example, wooo!

Formulation in 2D ( A) J This is the so-called div-grad equation Poisson s equation if ν is constant Appears in wildly diverse problems Thermal problem Electric potential Magnetic scalar potential Will take a look on these later Boundary conditions have different meanings in different problems

Boundary conditions = have to know something about the solution on the outer boundary Three common conditions for the vector potential formulation Dirichlet condition = A is constant = field is parallel to boundary Homogeneous Neumann condition = ν A perpendicular to boundary (alt. notation A n n = 0 = field is = A n) (Anti)periodic boundary = A 1 = ±A 2 on some parts of the boundary = used for modelling symmetry sectors

Interface conditions Interfaces (= boundaries wholly inside problem domain) are automatically included 1 H The conditions H t1 = H t2 and B n1 = B n2 are consequence of the Maxwells s equations Since we are approx. solving the equations, these conditions are also approx. satisfied B Will take a closer look later during the course, in the exercises B = μh

Example problem Magnetic field in the air gap and slots of a machine; = 0 (constant) J J J Nice to know = since ν is constant, the div-grad equation is reduced to the Poisson s equation ν 0 A = J ν 0 A = J 2 A = μ 0 J

Boundary conditions in the example The permeability of iron is much larger than those of the air and conductors On the air-iron interface Air Fe 1 Air 1 Fe Air Fe H 0 t Ht => Bt Bt => Bt Bt 0 0 Fe Fe Flux is almost perpendicular to the iron surface (Homogeneous) Neumann boundary condition ν A n = 0

Boundary conditions in the example Solution region can be reduced because of symmetry A 0 2 A 0 n A 0 Neumann condition on the right boundary No flux goes through left boundary. Dirichlet boundary condition A is constant: often A = 0 A n 0 2 A J 0 A n 0

Solution by finite element method 1. The problem domain is divided into small triangles called elements shape functions are generated based on these 2. Boundary conditions are applied 3. Computer solves the problem and draws nice pics for ya 4. Profit!

Assignment 1 Levitation melting http://i.kinja-img.com/gawkermedia/image/upload/t_original/ht8kqqbjsz6uns9q7eq b.gif Simulate this in FEMM Axisymmetric geometry Harmonic = sinusoidal current supply

Today we Conclusion 1. Understood the basic idea of FEM 2. Introduced the vector potential 3. and used it to derive the div-grad equation from the Maxwell s equations 4. Learned the most typical boundary conditions