Let the order of corrections to the unknowns be according to stage number, which in terms of the corresponding unknowns is
where X = [v1j, V2,j, . . . Vc.j, Tj, i^, ^y, . . . (.j (13-97)
Let the order of the linearized MEH functions also be according to stage number, which in terms of the corresponding nonlinear functions is
where Fj = [H, M1,j, M2-j, . . . Mc.j, E1j, . . . ECJ]T (13-99)
Corrections to unknowns for the kth iteration are obtained from
The next approximations to the unknowns are obtained from
where t is a damping (0 < t < 1) or acceleration (t > 1) factor. By ordering the corrections to the unknowns and the linearized functions in this manner, the resulting Jacobian of partial derivatives of all functions with respect to all unknowns is of a very convenient sparse matrix form of block tridiagonal structure.
Was this article helpful?