Page 520 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
        P. 520
     490                SLENDER STRUCTURES AND AXL4L FLOW
                                                                                          (F.29)
                     Consequently, the inverse operator zil exists  if  and  only if  Ak.i # 0. Introducing the
                      subset xk of Hk representing the eigenvectors of  YA with nonzero eigenvalue, it can be
                      said that the components of  fk(y) lying in Xk can be eliminated by  a proper choice of
                     hl(y), while the components of fk(y) not lying in Xk cannot be eliminated. In other words,
                      the terms of fk(y) lying in the range of 2~ can be eliminated, while those lying in the
                      kernel of  YA have to stay.
                        From the analysis, three important characteristics become apparent: (i) the normal form
                      method is local, since the coordinate transformation is generated in the neighbourhood of a
                      known solution (usually a fixed point for vector fields); (ii) the coordinate transformation
                      is  a nonlinear function of  the  dependent variables, but  it is  found by  solving a  linear
                      problem;  (iii) the  structure of  the  normal  form  is  determined entirely by  A,  since the
                      transformation depends only on the eigenvalues of A - see equation (F.29).
                        Examples of how to find normal forms may be found in many books, e.g. Guckenheimer
                      & Holmes (1983), Wiggins (1990), Arrowsmith & Place (1990). Because of its importance,
                      here we consider the case of a two-dimensional system with purely imaginary eigenvalues:
                                                                                          (F.30)
                      and, because of  its simplicity, we follow the methodology developed by Nayfeh (1993).
                      Equation (F.30) is first transformed into a single complex-valued equation using the trans-
                      formation
                                              x1  = c - 4,   x2  = i(C - f),              (F.31)
                      to obtain
                                 4 = it + ic [(a, - ia5)(< + 5)’ + i(a2 - ia6>(< + f)’(~ - 5)
                                     -(a3  - ia7)(~ +   - 5)’ - i(a4 - ia8)(< - 513] .    (F.32)
                      Using the methodology described previously, we assume
                                       2‘  = 9 + ~h(q, 3)   and   q = ir] + cg(q, 7).     (F.33)
                      Substituting (F.33) into (F.32) and equating the coefficients of E on both sides yields
                      Next, the function h must be chosen so as to eliminate the nonresonance terms, Le.  all
                      nonlinear terms that do not produce secular terms (which implies that  resonance terms
                      are  those producing secular terms). The form of equation (F.34) suggests choosing h in
                      the form
                                            h = r,q3 + r2q27 + r3qij2 + r4ij3.             (F.35)
     	
