【正文】 both sensitive and rapid in order to minimise the stress imposed on the equipment during the fault period (electrodynamic and thermal stresses). It is also essential that they should be selective, that is capable of eliminating only the faulty element whilst maintaining the supply to healthy parts of the work. Choice of the time/current characteristic Overcurrent protection relays are mainly characterised by their time/current characteristic. Several types are available: ? Independent or definite time relays (whose operating time is independent of current level) ? Dependent or inverse time relays (whose operating time depends on the current level). This last type of relay may be subdivided into three categories according to the IEC standard 2554: ? Inverse time relays (ITG7200 series ) ? Very inverse time relays (ITG7300 series ) ? Extremely inverse time relays (ITG7400 series ) ? Digital multicurve overcurrent relay (RMS700, RMS7000, RMST7000 and RMSA7000 series) No particular criterion exists for the systematic choice of one or other type of relay. However, dependent time relaying is preferable in the following cases where: ? The operation of the work includes highlevel shorttime overloads, ? Magising inrush currents at switchon may be considerable for several tenths of a second, ? Relay operation must be coordinated with a large number of fuses. On the other hand, independent time relays are preferable when the shortcircuit level is very high, or when it is likely to vary widely at a given point (for example when a work is supplied from small generators whose shortcircuit decrement curve falls off rapidly). As a basic rule, however, there is a general tendency to use independent time relays in Continental Europe and dependent time in AngloSaxon countries. Timedelayed overcurrent relays lend themselves to chronological selectivity. This however is not without a certain inconvenience, in that fault clearance time increases closest to the source, where shortcircuit levels are the highest. It is thus necessary to minimise the grading interval . The grading interval normally used for electronic protection relays is 400 ms, which is obtained by summating the following: ? Breaker fault clearance time, ? Summation of the time errors of the two relays, ? Overshoot of the upstream relay, ? Safety margin of approximately 100 ms. When transformer feeder protection is being considered it is advisable to use relays having an instantaneous highset unit. This is set above the secondary shortcircuit level and transformer inrush current, which then allows the operating time of upstream relays to be reduced thereby reducing the cable shortcircuit withstand requirements. The grading interval may be reduced if required, for times less than about 1 sec, to 250 ms, using relays with highstability timer circuits (which is the case with ITG 7000 or RMS 700/7000 relays). Accelerated selective protection In cases where the number of grading steps at one voltage level would lead to fault clearance times which are either too long for the withstand of the equipment in the work, or inpatible with the time allowed by the power distribution authority, it is possible to use an accelerated system. This consists of reducing the relay timedelay to a predetermined minimum value when the fault is on the section immediately downstream. The information required to achieve this is transmitted from the downstream relay to the upstream relay via