Selective coordination has one job: when a fault happens, the overcurrent device closest to it opens, and nothing upstream moves. The fault on the branch circuit stays a branch circuit problem. When coordination fails, the fault walks upstream — a shorted ballast trips a feeder, or a feeder fault drops the main — and loads that had nothing to do with the problem go dark with it.
How is selective coordination verified?
On paper, with time-current curves. Every protective device has a characteristic curve — how fast it operates at each level of current — plotted on log-log axes, current across the bottom, time up the side. Stack every device in a path on the same plot: the branch breaker, the feeder breaker, the main, the upstream fuse. Selectivity means the downstream device's curve clears the upstream device's curve — no overlap — across the entire range of fault current actually available at the downstream location.
That last clause is where the work is. It is easy to be selective at overload levels, where curves are spread apart by seconds. The hard region is high fault current and short time, where everything wants to operate in a few cycles. And the range that matters is not generic — it comes from a fault current calculation at each bus, which is why a coordination study is built on top of a short circuit analysis rather than run in isolation.
Where does the NEC require it?
Three systems carry a hard requirement:
- Emergency systems — Article 700. Egress lighting, exit signs, fire alarm loads. The code's logic is blunt: a fault in one emergency load must not take out the rest of the emergency system while the building evacuates.
- Legally required standby systems — Article 701. Loads the jurisdiction requires to have backup — smoke control, some elevators — with the same selective coordination obligation.
- Critical operations power systems — Article 708. Designated critical infrastructure facilities.
Healthcare facilities inherit the requirement through Article 517: the essential electrical system's branches are supplied through transfer switches and classified in a way that pulls in the emergency-system rules. (We cover the healthcare case in detail in Selective Coordination for Hospitals and Emergency Power Systems.)
The precise section numbers and allowances have shifted between code cycles — including how devices in series with no load tapped between them are treated — so on a real project, the governing text is whichever NEC edition your AHJ enforces. That is a detail worth settling at scoping, not at plan review.
Why is this harder than it looks?
Two reasons, and they are the two most common findings in our coordination work.
Instantaneous elements don't negotiate. A standard molded-case breaker's instantaneous trip operates in roughly a cycle once current crosses its threshold, regardless of what is downstream. Put two of them in series where the available fault current exceeds both instantaneous pickups, and both can unlatch on the same fault — which device clears first is a race, not a design. Published curves alone cannot resolve this region; manufacturers' tested series combinations and coordination tables, electronic trip units with short-time delay, or zone-selective interlocking are the tools that can.
Coordination and arc flash pull on the same settings. Slower upstream clearing buys coordination margin. Faster clearing lowers incident energy. The same short-time delay that keeps the main closed during a feeder fault also holds an arcing fault on the bus longer, and the arc flash study will price that decision in cal/cm² on the label. Tuning one without checking the other is how a facility ends up either cascading outages or hazard labels nobody wants to stand in front of. This is the strongest practical argument for running both studies on one model.
What does a coordination study actually deliver?
Ours produces the time-current curve plots for every path studied, a settings sheet for every adjustable device — pickup, delay, instantaneous, and the as-found versus recommended values — and a list of the spots where selectivity cannot be achieved with the installed equipment, with options priced against each other: different trip units, zone-selective interlocking, fuse substitutions, or accepting a documented non-selective pair where the code allows it.
It also fixes the everyday complaint that brings most people to us: nuisance tripping. The main that drops when the chiller starts, the feeder that trips on transformer inrush — those are usually pickup settings set too low for legitimate inrush, and the curves show it plainly.
The selective coordination service page covers scope, turnaround, and what it costs to run alongside an arc flash study. If you only remember one thing: selectivity is verified with curves and tables against calculated fault current, never assumed from breaker sizes — ampere ratings in a ratio tell you nothing about what happens in the first cycle of a fault.