16.
21 Techniques of Structural Analysis and
Design
Spring 2005
Unit #9 Calculus of Variations
Ra ul Radovitzky March 28, 2005
Let u be the actual conguration of a structure or mechanical system. u satises the displacement boundary conditions: u = u on Su . Dene: u = u + v, where:
: scalar
v : arbitrary function such that v = 0 on Su
We are going to dene v as u, the rst variation of u: u = v Schematically:
u1 u(b) u(a) u2
(1)
v a b
As a rst property of the rst variation :
du du dv = + dx dx dx
dv so we can identify dx with the rst variation of the derivative of u: du dv = dx dx But: dv dv d (u) = = dx dx dx We conclude that: du d = (u) dx dx
Consider a function of the following form: F = F (x, u(x), u (x)) It depends on an independent variable x, another function of x (u(x)) and its derivative (u (x)). Consider the change in F , when u (therefore u ) changes: F = F (x, u + u, u + u ) F (x, u, u ) = F (x, u + v, u + v ) F (x, u, u ) expanding in Taylor series: F F 1 2F 1 2F 2 v + v + ( v ) + (v )(v ) + F 2 u u 2! u 2! uu F F = v + v + h.o.t. u u First total variation of F: F F = lim 0 F (x, u + v, u + v ) F (x, u, u ) = lim 0 F v + F v F F u = v + v = lim u 0 u u F F F = u + u u u F = F + 2
Note that: dF (x, u + v, u + v )
d =0 since: F (x, u + v, u + v ) F (x, u + v, u + v ) dF (x, u + v, u + v ) = v+ v d u u evaluated at = 0 dF (x, u + v, u + v ) F (x, u, u ) F (x, u, u ) v+ v
=
d u u =0 F = Note analogy with dierential calculus. (aF1 + bF2 ) = aF1 + bF2 linearity (F1 F2 ) = F1 F2 + F1 F2 etc The conclusions for F (x, u, u ) can be generalized to functions of several ui : independent variables xi and functions ui , x j
ui
F xi , ui , xj We will be making intensive use of these properties of the variational operator :
du
d d dv (u) = (v ) = =
dx
dx
dx
dx
udx = vdx = vdx = udx Concept of a functional
b I (u) =
a
F (x, u(x), u (x))dx
First variation of a functional: I = F (x, u(x), u (x))dx = F (x, u(x), u (x)) dx F F I = u + u dx u u Extremum of a functional u0 is the minumum of a functional if: I (u) I (u0 )u A necessary condition for a functional to attain an extremum at u0 is: I (u0 ) = 0, or dI (u0 + v, u0 + v ) =0 d =0
Note analogy with dierential calculus. Also dierence since here we require dF = 0 at = 0. d b F F I = u + u dx u u a Integrate by parts the second term to get rid of u . b F d F d F I = u + u u dx u dx u dx u a b F F d F b = udx + u u dx u u a a Require u to satisfy homogeneous displacement boundary conditions: u(b) = u(a) = 0 Then: b F d F I = udx = 0, u dx u a 4
u that satisfy the appropriate dierentiability conditions and the homoge neous essential boundary conditions. Then: F d F =0 u dx u These are the EulerLagrange equations corresponding to the variational problem of nding an extremum of the functional I . Natural and essential boundary conditions A weaker condition on u also allows to obtain the Euler equations, we just need: F
b u
= 0 u a which is satised if: u(a) = 0 and u(b) = 0 as before u(b) = 0 and
F
(a) u F (a) u F (b) u
=0
= 0 and u(b) = 0 = 0 and
F (b) u
=0
= 0, or u = u0 on Su Essential boundary conditions: u
Su F Natural boundary conditions: u = 0 on S . Example: Derive Eulers equation corresponding to the total po tential energy functional = U + V of an elastic bar of length L, Youngs modulus E, area of cross section A xed at one end and subject to a load P at the other end. L EA
du
2 (u) = dx P u(L) 2 dx 0 Compute the rst variation:
EA du
du
2 = dx P u(L) 2 dx dx
Integrate by parts d du d du = EA u EA u dx P u(L) dx dx dx dx L du d du L = u EA dx + EA u P u(L) dx dx dx 0 0 Setting = 0, u / u(0) = 0: d du EA =0 dx dx du P = EA dx L
Extension to more dimensions
F F ui,j dV I = ui + ui,j V ui F F F ui + ui ui dV = xj ui,j xj ui,j V ui Using divergence theorem: F F F ui nj dS I = ui dV + xj ui,j V ui S ui,j Extremum of functional I is obtained when I = 0, or when: F F = 0 , and ui xj ui,j ui = 0 on Su F nj onS Su = St ui,j The boxed expressions constitute the EulerLagrange equations correspond ing to the variational problem of nding an extremum of the functional I . 6
I= F (xi , ui , ui,j )dV
V
