SynopsisA method of computing switching overvoltages based on the numerical inversion of the modified Fourier transform is discussed. Several applications are made to practical studies of varying degrees of complexity. The results of a 3-phase switching problem are shown to be in agreement with those obtained by other workers using analogue simulation. The method allows more complete representation of the problem than was previously possible. Indeed, during the course of the work, the type of data necessary to permit an even closer representation became apparent, and attention is drawn to the desirability of recording this on future site tests.
List of symbolst = time T = travel time a -'smoothing' parameter on = angular frequency used as transform parameter f+(a + jco) = modified Fourier transform of/(/) Cl = integration range in the truncated inversion integral a = sigma factor / = length of line m = integer transform parameter associated with finite cosine transform p = Laplace-transform parameter v, V, v, V (or with suffixes) /, I, /, / (or with suffixes) Z, 2 (or with suffixes) L,M rs = voltages and their transforms = currents and their transforms = impedances and their transforms mutual induc-1inductance per unit length tance per unit length L s = source-side inductance R = resistance per unit length R s = source-side resistance r = switching resistor C rs , K r = capacitance per unit length a 2 = {R + (a+joj)L}(a + jw)C
The Modified Fourier TransforrnUse has been made of the inverse Fourier transform viz. 00 f(t) = _L JJ(oo)exp (joot) doo 2,.-00 whereJ( 00) is the frequency response and the integration is carried out along the real axis, in the approach of Lego and Sze. (2) For numerical integration the range must be made finite and any resulting Gibbs oscillations can be reduced by incorporating the sigma factor described in Part I. However, the poles of J(00) may, and in many cases do, lie close to the real axis. This causes the integrand to peak over a series of small intervals lying in the neighbourhood of the poles and hence for accurate numerical integration it is necessary to take a very small step length. This limitation can be overcome by using the modified Fourier transformtti and at the same time this secures the additional advantage of avoiding problems associated with the convergence of the integrals involved. The method Lego and Sze devised to overcome this latter difficulty was to introduce an artificial forcing function they denoted by U(a, t) (see Ref. 2, p. 1032, equation 7).In the present approach we consider the problem of finding the response of a linear system to some input starting at time zero. Suppose that the differential equation of the system is known and can be written in the formwheref(t) is the output, g(t) the input both of which are zero for t < 0. In taking the modified Fourier transform equation (10) is multiplied throughout by exp {-(a +joo) t} and integrated w.r.t, tover the range (-00,00), or,sincef(t),g(t) arezerofort < 0, over the range (0, 00); equation (10) then becomes F(a + joo)J+(a + joo) = K+(a + joo)at The University of Iowa Libraries on June 4, 2016 ije.sagepub.com Downloaded from
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.