ce 257 discrete methods of structural analysis · 2020. 6. 30. · ce 257 5. introduction •thus,...

67
CE 257 Discrete Methods of Structural Analysis 1 This lecture material is not for sharing/distribution outside of CE257 and CE297 class.

Upload: others

Post on 24-Jan-2021

0 views

Category:

Documents


0 download

TRANSCRIPT

  • CE 257 Discrete Methods of Structural Analysis

    1

    This lecture material is not for sharing/distribution outside of CE257 and CE297 class.

  • Last meeting

    • Time-dependent Problems

    • 2D Triangular Elements

    2CE 257

  • For this meeting

    • Single-variable problems in two dimensions

    Reference for this lecture:

    J. N. Reddy, An Introduction to the Finite Element Method (Chapter 8)

    3CE 257

  • CE28 Analytical and Computational Methods in CEIII

    4

    Introduction

  • Introduction

    • FE analysis of two dimensional (2D) problemsinvolve the same basic steps as those describedfor one-dimensional problems.

    • The analysis is somewhat complicated by the factthat 2D problems are described by PDEs overgeometrically complex regions. The boundary Γ ofa two-dimensional domain Ω is, in general, acurve. Therefore, finite elements are simple twodimensional geometric shapes that allowapproximations of a given two-dimensionaldomain as well as the solution over it

    5CE 257

  • Introduction

    • Thus, in 2D problems we not only seek anapproximate solution to a given problem on adomain, but we also approximate the domain by asuitable finite element mesh.

    • Consequently, we will have approximation errorsdue to the approximation solution as well asdiscretization errors due to the approximation ofthe domain in the FE analysis of 2D problems.

    6CE 257

  • Introduction

    • FE mesh: triangles, rectangles/quadrilaterals

    7CE 257

  • CE28 Analytical and Computational Methods in CEIII

    8

    Boundary Value Problems

  • Boundary Value Problems

    The model equation:

    • Consider finding the solution 𝑢(𝑥, 𝑦) of the2𝑛𝑑 order PDE

    • for given data 𝑎𝑖𝑗 , 𝑎00, and 𝑓, and specified BCs.

    9CE 257

  • Boundary Value Problems

    • As a special case, we can obtain the Poissonequation from the model DE by setting 𝑎11 =𝑎22 = 𝑘(𝑥, 𝑦) and 𝑎12 = 𝑎21 = 0

    • where ∇ is the gradient operator.

    10CE 257

  • Boundary Value Problems

    In Cartesian coordinate system:

    11CE 257

  • Boundary Value Problems

    Major steps in developing the FE model of model DE:

    12CE 257

  • CE28 Analytical and Computational Methods in CEIII

    13

    F. E. Discretization

  • FE Discretization

    • The representation of a given region by a set ofelements (i.e. discretization or mesh generation) is animportant step in FE analysis.

    • The choice of element type, number of elements, anddensity of elements, depends on the geometry of thedomain, the problem to be analyzed, and the degreeof accuracy desired.

    • In general, the analyst is guided by his or her technicalbackground, insight into the physics of the problembeing modeled (e.g. qualitative understanding of thesolution) and experience with the FE modeling.

    14CE 257

  • FE Discretization

    • The general rules of mesh generation for FEformulation include:

    15CE 257

  • CE28 Analytical and Computational Methods in CEIII

    16

    Weak Form

  • Weak Form

    • First step is to multiply the DE with a weightfunction 𝑤, which is assumed to be differentiableonce with respect to 𝑥 and 𝑦 and then integratethe equation over the element domain Ω𝑒 .

    • where

    17CE 257

  • Weak Form

    • Second step: IBP to distribute the differentiationamong 𝑢 and 𝑤 equally.

    • We will use the following identities:

    18CE 257

  • Weak Form

    • Next we use the component form of the gradient(or divergence) theorem

    • where 𝑛𝑥 and 𝑛𝑦 are the components (i.e. the

    direction cosines) of the unit normal vector

    19CE 257

  • Weak Form

    • We obtain

    Let 𝑞𝑛 be boundary expression:

    20CE 257

  • Weak Form

    • By definition, 𝑞𝑛 is taken positive outward fromthe surface as we move counterclockwise alongthe boundary Γ𝑒.

    21CE 257

  • Weak Form

    22CE 257

  • Weak Form

    • Rewriting the weak form:

    23CE 257

  • CE28 Analytical and Computational Methods in CEIII

    24

    Finite Element Model

  • Finite Element Model

    • The weak form requires that the approximationchosen for 𝑢 should be atleast linear in both 𝑥 and𝑦 so that there are no terms that are identicallyzero.

    • Since the primary variable is simply the functionitself, the Lagrange family of interpolationfunctions is admissible.

    • Approximation of 𝑢 over the typical finite elementΩ𝑒:

    25CE 257

  • Finite Element Model

    • where

    Lagrange interpolation property

    26CE 257

  • Finite Element Model

    • Substituting the FE approximation for 𝑢 into theweak form, we obtain

    27CE 257

  • Finite Element Model

    28CE 257

    or

    where

  • Finite Element Model

    • In matrix notation

    • where

    29CE 257

  • CE28 Analytical and Computational Methods in CEIII

    30

    Interpolation Functions

  • Derivation of Interpolation functions

    • The FE approximation 𝑢ℎ𝑒(𝑥, 𝑦) over an element

    Ω𝑒 must satisfy the following conditions in orderfor the approximate solution to converge to thetrue solution:

    1. 𝑢ℎ𝑒 must be continuous as required in the weak

    form of the problem (i.e. all terms in the weak formare represented as nonzero values)

    2. The polynomial used to represent 𝑢ℎ𝑒 must be

    complete (i.e. all terms, beginning with a constantterm up to the highest order used in the polynomial)

    31CE 257

  • Derivation of Interpolation functions

    • The FE approximation 𝑢ℎ𝑒(𝑥, 𝑦) over an element

    Ω𝑒 must satisfy the following conditions in orderfor the approximate solution to converge to thetrue solution:

    3. All terms in the polynomial should be linearlyindependent

    32CE 257

  • Triangular element

    • Complete linear polynomial in 𝑥 and 𝑦 in Ω𝑒

    • We can denote:

    33CE 257

  • Triangular element

    • Approximation:

    34CE 257

  • Triangular element

    35CE 257

  • Triangular element

    • Shapes to avoid:

    36CE 257

  • Triangular element

    • inverting coefficient matrix

    37CE 257

  • Triangular element

    • Substituting 𝑐𝑖 back to the approximation for 𝑢 ∶

    • where:

    38CE 257

  • Triangular element

    • Interpolation functions:

    39CE 257

  • Triangular element

    • Properties of interpolation functions:

    40CE 257

  • Triangular element

    • Representation of continuous function

    41CE 257

  • Linear Rectangular Element

    • Consider the complete polynomial

    • which contains four linearly independent termsand is linear in 𝑥 and 𝑦, with a bilinear term in 𝑥and 𝑦. This polynomial requires an element withfour nodes.

    42CE 257

  • Linear Rectangular Element

    • Incompatible four node triangular elements

    43CE 257

  • Linear Rectangular Element

    • Using a rectangular element with sides 𝑎 and 𝑏,we choose a local coordinate system, ( ҧ𝑥, ത𝑦) toderive the interpolation functions.

    44CE 257

  • Linear Rectangular Element

    45CE 257

  • Linear Rectangular Element

    46CE 257

    Solving for 𝑐𝑖:

  • Linear Rectangular Element

    • Simplifying..

    • where

    47CE 257

  • Linear Rectangular Element

    • Interpolation functions:

    48CE 257

  • Linear Rectangular Element

    • Properties of interpolation functions:

    49CE 257

  • Linear Rectangular Element

    • Quadratic Elements (personal reading)

    50CE 257

  • CE28 Analytical and Computational Methods in CEIII

    51

    Evaluation of Matrices and Vectors

  • Evaluation of Element Matrices and Vectors

    • We rewrite [𝐾𝑒] as the sum of the five basicmatrices

    52CE 257

  • Evaluation of Element Matrices and Vectors

    • Element Matrices of a Linear Triangular Element(for a Poisson problem)

    53CE 257

  • Evaluation of Element Matrices and Vectors

    • Element Matrices of a Linear Rectangular Element

    54CE 257

  • Evaluation of Element Matrices and Vectors

    • Element Matrices of a Linear Rectangular Element

    55CE 257

  • Evaluation of Element Matrices andVectors

    56CE 257

  • Evaluation of Boundary Integrals

    • Boundary Integral

    57CE 257

  • CE28 Analytical and Computational Methods in CEIII

    58

    Assembly of Element Equations

  • Assembly of Element Equations

    59CE 257

  • Assembly of Element Equations

    • In matrix form:

    60CE 257

  • CE28 Analytical and Computational Methods in CEIII

    61

    Postcomputations

  • Postcomputations

    • The finite element solution at any point (𝑥, 𝑦) inan element delta is given by

    • and it derivatives are computed as

    62CE 257

  • Postcomputations

    • The derivatives will not be continuous atinterelement boundaries because continuity ofderivatives is not imposed during the assemblyprocedure.

    • The weak form of the equations suggests that theprimary variable is 𝑢, which is to be carried as thenodal variable.

    63CE 257

  • CE28 Analytical and Computational Methods in CEIII

    64

  • Finite Element Method

    References:

    1. J. Fish, T. Belytschko, A First Course in Finite Elements

    2. J. N. Reddy, An Introduction to the Finite Element Method

    3. O.C. Zienkiewicz, R.L. Taylor, J.Z. Zhu, The Finite Element Method, Its Basis & Fundamentals

    4. K-J. Bathe, Finite Element Procedures

    5. K. Leet, C-M., Uang, A.M. Gilbert, Fundamentals of Structural Analysis

    65CE 257

  • This lecture

    • Single-variable problems in two dimensions

    Reference for this lecture:

    J. N. Reddy, An Introduction to the Finite Element Method (Chapter 8)

    66CE 257

    This lecture material is not for sharing/distribution outside of CE257 and CE297 class.

  • CE28 Analytical and Computational Methods in CEIII

    67