Link Click
Link Click
Introduction
These methods involve replacing the derivatives in a given ordinary dierential equation with equivalent dierence approximations. It follows that any convenient nite dierence can be used in approximating the solution.
CentralDierence Method
The rst and second derivative (see Fig. 1) may be approximated by using a central dierence formula: xi = xi+1 xi1 2h xi = xi+1 2xi + xi1 , h2 (1)
x(t)
Dt = h
x i+1
xi x i-1
i-1
i+1
2 CENTRALDIFFERENCE METHOD
The governing equation of a viscously damped single degree of freedom is m d2 x dx t + c + kx = F (t) = F0 1 sin 2 dt dt 2t0 (2)
To obtain a satisfactory solution, we must select a time step t that is smaller than a critical time step tcri . Let the initial conditions be given by x (t = 0) = x0 and x (t = 0) = x0 . Replacing the derivatives by the central dierences and writing at grid point i gives x (t) xi m xi+1 2xi + xi1 t2 +c F (t) Fi xi+1 xi1 + kxi = Fi 2t
(3)
Solution of Eq. (3) for xi+1 yields the following recurrence formula: 2m c m k xi + 2 xi1 + Fi 2 t 2t t , m c + t2 2t
xi+1 =
(4)
with i = 0, 1, 2, . . .. Recurrence formula permits to calculate the displacement of the mass if we know the previous history of displacements at time ti and ti1 as well as the present external force Fi . Repetead application of Eq. (4) yields the complete time history of the behavior of the system. Note that the solution of xi+1 is based on the use of the equilibrium equation at time ti . For this reason, this integration procedure is called an explicit integration method. We need to nd the value of x1 by using the Eqs. (1). By substituting the known values of x0 and x0 into Eq. (2), x0 can be found: x0 = 1 [F (t = 0) cx0 kx0 ] , m (5)
Application of Eqs. (1) at i = 0 yields the values of x1 : x0 = x1 x1 x1 x1 = 2t x0 2t x1 2x0 + x1 x0 = x1 + x1 = 2x0 + x0 t2 t2 x1 + x1 (x1 x1 ) = 2x0 + x0 t2 2t x0 2 t x1 = x0 tx0 + x0 . 2
(6)
Subtracting
(7)
2 CENTRALDIFFERENCE METHOD
Numerical method require a time step t smaller than a critical time step tcri : t < tcri = n (8)