Discrimination & Selectivity: Why the Upstream Device Shouldn't Trip

A fault on one lighting circuit should trip that circuit's breaker — not the main incomer that feeds the whole building. Keeping the fault contained to the device nearest to it is called discrimination (or selectivity). It sounds automatic. It isn't — it's a design outcome you have to check, and it's routinely assumed rather than proven.

What Discrimination Means

When a fault occurs, only the protective device immediately upstream of the fault should operate; every device further back stays closed. Achieve that across the full range of currents — from a small overload up to the maximum prospective fault — and you have total discrimination. Achieve it only up to some current below the fault level and you have partial discrimination, with a black-out risk above that point.

The Picture: Two Curves That Mustn't Cross

Every protective device has a time–current characteristic: the bigger the current, the faster it operates. Overlay the downstream device's curve and the upstream device's curve on the same axes. For discrimination, the upstream curve must sit clear above and to the right of the downstream curve, all the way out to the prospective fault current. Where they touch or cross, discrimination is lost.

Two Regions, Two Problems

• Overload region (low current, long time). Here discrimination is about a time gap — the upstream device is slower for the same current. Usually straightforward with a sensible rating step.
• Fault region (high current, short time). Here both devices are trying to operate almost instantly, so a time gap alone isn't enough — you're relying on current thresholds and let-through energy (I²t). This is where discrimination quietly fails.

Why Fuses and MCBs Behave Differently

Fuses discriminate well on a simple ratio — roughly 1.6:1 between ratings is often enough, because their let-through energy scales predictably. MCBs are harder: at high fault currents their magnetic (instantaneous) trips overlap, so two MCBs in series may both trip regardless of rating. That's why serious distribution boards often use MCCBs or ACBs with adjustable short-time settings, or zone-selective interlocking (ZSI), to force selectivity at the fault end.

Where It Goes Wrong

• Assuming a rating ratio guarantees discrimination without checking the fault-current end.
• Mixing manufacturers and relying on generic curves instead of tested discrimination tables.
• Overlooking MCB type (B/C/D) — the instantaneous bands overlap more than people expect.
• Using a system-wide fault level instead of the actual prospective fault current at each board.

How to Prove It

Discrimination is demonstrated with the manufacturer's discrimination tables and by overlaying the real device curves against the prospective fault current calculated at each board — not a nominal figure. That's part of a protection coordination study: the fault levels, the device curves and the discrimination check in one model, presented so an approving engineer can audit it.