New lightning arrester standard

Abstract

3.3 seconds. This result, with the calculated value 12 per cent higher than the measured value, is typical of the time constants that have been checked. The error has generally been of the order of 10 or 15 per cent, with the calculated result being high. It should be emphasized that the foregoing results were computed by starting with the a-c saturation curve of the core, and that no experi mental corrections were used at any stage of calculations.If desired, the path of a transient that spans a markedly nonlinear portion of the flux-versus-current graph can be computed graphically by a step-by-step method. The main error here is caused by the uncertainty introduced into the initial condition by the hysteresis effect.If a comparatively large control voltage is used to force the control current to change rapidly to a point well above the knee of the control-circuit magnetization curve, the i m R e drop can be neglected during most of the change in flux, and the time required for the flux linkages to change an amount Αψ <: is only slightly greater than A\p e /E c , where Έ ϋ is the suddenly-applied control voltage. For example, reactor 66G912 connected in the self-feedback circuit was started at ψ £ = 6 weber-turns (I ae =0.8 ampere), and an electromotive force of 47 volts was suddenly applied to the control winding. This was sufficient to produce a steadystate control current of 300 milliamperes, which is well above the knee of the curve of Figure 11. The final value of φ £ is, from Figure 11, about 21 weber-turns (7 ER =10.2 amperes). Therefore, the approximate time of rise is A^c/jE'C=(21-6)/47 = 0.32second.An oscillogram showed that the load current rose almost linearly to 10 amperes in 0.31 second.