Electrical Protection Coordination: How a Relay Can Trip Twice for the Same Fault (and How to Prevent It)
Why can a relay trip twice for the same fault, even when its coordination "looks fine on paper"? It's one of the questions clients ask us most, and the answer almost always comes down to a detail a poorly verified selectivity study lets slip through. In this article — and the video that goes with it — we audit and correct a real case using RTC2 (Relay Tripping Curves PRO2), applying the IEEE Std 242 criteria (the "Buff Book"). This isn't just a case story: it's also a practical guide to verifying coordination in any protection chain.
The real case
A client reported that their Relay 52 (13.2 kV) tripped twice for a single three-phase fault that occurred downstream, at Bus 1A (480 V). On paper, the protection chain — CB1A, CB1, and Relay 52 — had its curves plotted and its coordination time intervals (ΔT) looked reasonable. Yet in the real event, two devices tripped almost simultaneously for the same fault — something a properly done selectivity study should never allow.
What is the ImaxF criterion, and why is it the foundation of everything?
Before plotting any curve, you have to define which current the coordination between two consecutive devices will be checked against. That reference current is the maximum fault current (ImaxF), and IEEE Std 242 defines three different ways to calculate it, depending on which pair of devices is being coordinated:
Breaker–Breaker (CB–CB): ImaxF = the minimum between the downstream CB's first non-inverse band pickup and the maximum short-circuit current at that node.
Breaker–Relay, through a transformer (CB–Relay): ImaxF = the minimum between the CB's instantaneous (or short-time) pickup, referred to the high-voltage side, and the maximum short-circuit current at the low-voltage bus, also referred across the transformer.
Relay–Relay: ImaxF = the minimum between the downstream relay's instantaneous pickup and the maximum short-circuit current at that node.
In all three cases, the logic is the same: you take the lower value between two candidates, because that's the point up to which selectivity between the two devices genuinely needs to be guaranteed. Getting this wrong — or skipping it entirely — is the most common reason coordination "looks fine" on the plot but fails in reality.
The diagnosis: two causes, not one
Reviewing the case in depth with RTC2, we found the problem wasn't caused by one thing, but by two issues acting together:
Relay 52's original instantaneous (ANSI 50) setting was fixed at 674 A — too close to the fault current reflected on the 13.2 kV side (~625 A). That narrow margin meant the relay could trip on its instantaneous element for a fault that should have been left for CB1 to clear first.
CB1's instantaneous trip (its I3 setting) was fixed, non-adjustable, and calibrated below the real fault current at Bus 1A, which was 17,179 A. In other words, CB1 was tripping instantaneously for a current far lower than what it should actually have withstood before acting.
Both problems combined caused CB1A, CB1, and Relay 52 to trip almost simultaneously for the same fault — exactly what the client reported.
The fix
With the diagnosis clear, the fix involved recalculating and re-adjusting Relay 52's instantaneous setting so it fell outside the zone of unwanted tripping, and verifying that the coordination time interval (ΔT) between each pair of devices in the chain fell within the optimal 0.2 to 0.4 second range — short enough to avoid a simultaneous operation, but long enough not to expose the upstream device to unnecessary thermal stress. All of this is verified interactively in RTC2 without leaving the program: curves are plotted, the sweep cursor is used to read times directly off the plane, and the final selectivity report is generated.
The general lesson
This case illustrates something that applies to any installation: protection coordination isn't complete just because the curves are plotted and the ΔTs are calculated once. Instantaneous settings need to be verified with the same rigor as inverse-time curves, and currents must be correctly referred across a transformer when one is involved — because a single miscalibrated element can defeat coordination for the entire chain, no matter how well-tuned the rest of it is.
Full video of the case — now available with English audio (or Spanish narration with English subtitles):
Comments