Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 48 views

    ME5204 Finite Element Analysis: Class 7: Piecewise Linear Finite Elements and Properties of The Stiffness Matrix

    Uploaded by

    Obulapuram Sai Pavan Kumar me17b157
    The document discusses piecewise linear finite elements and properties of the stiffness matrix K. It describes: 1) The piecewise linear finite element space Vh is partitioned into n sub-intervals with nodes xA. Basis functions NA(x) are defined on each sub-interval that are linear and continuous. 2) The stiffness matrix K is symmetric and banded due to non-overlapping support of the basis functions. 3) K is positive definite, implying it has a unique inverse and positive eigenvalues, which is proven using the properties of the bilinear form a(u,v).

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    ME5204 Finite Element Analysis: Class 7: Piecewise Linear Finite Elements and Properties of The Stiffness Matrix

    Uploaded by

    Obulapuram Sai Pavan Kumar me17b157
    48 views13 pages
    The document discusses piecewise linear finite elements and properties of the stiffness matrix K. It describes: 1) The piecewise linear finite element space Vh is partitioned into n sub-intervals with nodes xA. Basis functions NA(x) are defined on each sub-interval that are linear and continuous. 2) The stiffness matrix K is symmetric and banded due to non-overlapping support of the basis functions. 3) K is positive definite, implying it has a unique inverse and positive eigenvalues, which is proven using the properties of the bilinear form a(u,v).

    Original Title

    Class7_FE_Class_Aug_2021(1)

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses piecewise linear finite elements and properties of the stiffness matrix K. It describes: 1) The piecewise linear finite element space Vh is partitioned into n sub-intervals with nodes xA. Basis functions NA(x) are defined on each sub-interval that are linear and continuous. 2) The stiffness matrix K is symmetric and banded due to non-overlapping support of the basis functions. 3) K is positive definite, implying it has a unique inverse and positive eigenvalues, which is proven using the properties of the bilinear form a(u,v).

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    48 views13 pages

    ME5204 Finite Element Analysis: Class 7: Piecewise Linear Finite Elements and Properties of The Stiffness Matrix

    Uploaded by

    Obulapuram Sai Pavan Kumar me17b157
    The document discusses piecewise linear finite elements and properties of the stiffness matrix K. It describes: 1) The piecewise linear finite element space Vh is partitioned into n sub-intervals with nodes xA. Basis functions NA(x) are defined on each sub-interval that are linear and continuous. 2) The stiffness matrix K is symmetric and banded due to non-overlapping support of the basis functions. 3) K is positive definite, implying it has a unique inverse and positive eigenvalues, which is proven using the properties of the bilinear form a(u,v).

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 13

    ME5204 Finite Element Analysis

    Class 7: Piecewise linear finite elements and properties of the


    stiffness matrix

    August 23, 2021


    Piecewise linear finite element space

    I The examples that we saw previously used basis functions


    which are cases of the picewise linear finite element space.
    I In the general case V h is n-dimensional. The domain [0,1] is
    partitioned into n non-overlapping sub-intervals.
    I The typical interval is denoted by [xA , xA+1 ], where xA < xA+1
    and A = 1, 2, .... n. Also x1 = 0 and xn+1 = 1. The xA ’s are
    called nodes. The sub-intervals are referred to as elements.
    I It is not required that the length of the elements be equal. The
    mesh parameter h is generally taken as the length of the largest
    interval.
    I The shape functions are defined as follows:
    For 2 ≤ A ≤ n


     x − xA−1

     , xA−1 ≤ x ≤ xA
     hA−1


    NA (x) = xA+1 − x (1)
     , xA ≤ x ≤ xA+1



     hA
    0, elsewhere

    I Whereas for the boundary nodes,

    x2 − x
    N1 (x) = , x1 ≤ x ≤ x2 (2)
    h1
    x − xn
    Nn+1 (x) = , xn ≤ x ≤ xn+1 (3)
    hn
    Figure: Basis functions for piece-wise linear finite element space
    I These shape functions are called.“hat”, or “roof” functions for
    obvious reasons.
    I It is noted that NA (xB ) = δAB , where δAB is the Kronecker
    delta i.e., δAB = 1 if A = B, and δAB = 0 if A 6= B.
    I To further clarify, NA takes the value 1 at xA and is zero at all
    other nodes.
    I A typical member wh ∈ V h has the form ΣnA=1 cA NA . It
    appears as shown in the figure below.
    Figure: A typical member wh ∈ V h
    I On each element w h is a linear polynomial in x.
    I It is instantly clear that w h has discontinuous slope across each
    element boundary.
    I It is noted that the essential boundary condition w h (1) = 0 is
    met.
    I It is also noted that wh is zero if and only if each cA = 0, A =
    1, 2, ....., n.
    I Typical members of δ h are obtained by adding gh = g Nn+1 to
    typical members of V h . This ensures that uh (1) = g.
    I The piece-wise linear finite element functions are the simplest
    and most widely used finite element functions for 1-D problems.
    Properties of K

    I It is noted that the shape functions are zero outside a


    neighborhood of node A. As a result many of the entries of K
    are zero.
    If B > A + 1,
    Z 1
    KAB = NA,x NB,x dx = 0. (4)
    0

    I Clearly by symmetry also true for A > B + 1.


    I Thus the non-zero entries are located in a band across the
    main diagonal. K is banded. Banded matrices deliver
    significant computational savings.
    Figure: If B>A+1, then the non-zero portions of NB and NA , do not
    overlap.
    Figure: banded structure of K
    Properties of K

    I A n × n matrix is called positive definite if

    cT Ac ≥ 0, for all n-vectors c and


    cT Ac = 0, implies c = 0. (5)

    I It maybe shown that a symmetric positive definite matrix


    possesses a unique inverse.
    I It is also noted that the eigenvalues of a positive definite
    matrix are real and positive.
    I Proposition The n × n stiffness matrix K is positive definite.
    I Proof:
    Let cA A = 1, 2, .... , n, be the components of c, an arbitrary
    vector. These cA ’s can be used to construct a member of V h ,
    wh = ΣnA=1 cA NA , where NA ’s are the basis functions of V h .
    Then,

    cT Kc = ΣnA,B=1 cA KAB cB (6)


    = ΣnA,B=1 cA a(NA , NB )cB (7)
    !
    = a ΣnA=1 cA NA , ΣnB=1 cB NB (8)

    = a(w h , w h ) (9)
    Z 1
    = (w,xh )2 dx (10)
    0
    ≥ 0 (11)

    I Assume cT Kc = 0. Now by previous result,


    Z 1
    (w,xh )2 dx = 0 (12)
    0

    Consequently w h must be a constant. Since wh ∈ V h , wh (1)


    = 0. Combining these facts, we conclude that wh (x) = 0 for
    all x ∈ [0,1]. This is possible only if cA = 0 for all A = 1, 2,
    ...., n. Thus c = 0.
    Hence proved.
    I In Summary:
    K is,
    a. Symmetric
    b. Banded
    c. Positive-definite

    You might also like