Fuse

What it is (precise)

Start with the two definitions everything else hangs off. IEC 60269-1 Ed. 5.0:2024 clause 3.2.2 defines a current-limiting fuse-link as one that, during and by its operation in a specified current range, limits the current to a substantially lower value than the peak of the prospective current.

Clause 3.2.3 defines a "g" fuse-link as full-range breaking capacity — formerly called general purpose — capable of breaking all currents which cause melting of the fuse element, up to its rated breaking capacity. Clause 3.2.4 defines an "a" fuse-link as partial-range, formerly back-up: it breaks all currents between the lowest current indicated on its operating time-current characteristic, k2·In in the standard's Figure 2, and its rated breaking capacity.

The band below k2·In is the whole distinction. Note 1 to clause 3.2.4 records that "a" fuse-links are generally used for short-circuit protection, and that where protection is required against overcurrents less than k2·In they are used in conjunction with another suitable switching device designed to interrupt those small overcurrents. The second letter names what is protected.

IEC 60269-4 Ed. 5.0:2009 clause 5.7.1 states that the first letter shall indicate the breaking range, and lists only R and S as its own second letters, for semiconductor protection; the general two-letter grammar sits in IEC 60269-1, clause 6.7.1 in Ed. 5.0. So gG reads as full-range general purpose, mainly cable protection, and aM as partial-range back-up for short-circuit protection of motor circuits — meanings taken from manufacturer training material, because the defining IEC clause is paywalled.

Operation splits into two stages with separate names, and conflating them is a real error. Pre-arcing time — melting time — is the interval between the beginning of a current large enough to cause a break in the fuse element and the instant an arc is initiated. Arcing time runs from arc initiation to final arc extinction. Operating time, also called total clearing time, is by definition the sum of the two.

Those are IEC vocabulary entries IEV 441-18-21, 441-17-37 and 441-18-22, reproduced here from the IEC 60282-2:2008 preview because the equivalent IEC 60269-1 clause is not in the free extract. I²t, the Joule integral, is the integral of the square of the current over a given interval: pre-arcing I²t is taken over the pre-arcing time, operating I²t over the whole operating time.

Why it matters in a real grid-scale project

DC is where fuse selection stops being a lookup. On AC the current crosses zero twice a cycle; on DC there is no natural voltage or current zero, so only the fuse's own arc forces the current down, and a large circuit time constant L/R makes that materially harder.

You therefore cannot pick a DC voltage rating from the circuit's working voltage — the manufacturer has to plot the curve L/R = f(U) from maximum-energy test results, and a DC rating is only meaningful as a triple: voltage, time constant, and the range of possible fault currents. IEC 60269-1 Clause 1 even recommends that, where later parts do not state DC performance, it be given in the manufacturer's literature. Take a 1500 V figure off a datasheet without the L/R it was earned at and you have half a number.

Fuse levels map onto the DC hierarchy: cells form a module, modules in series form a rack, racks in parallel form a section, sections in parallel form a container, and the container feeds the power conversion system.

Mersen's application guide maps faults to the device that clears them — a fault inside a module to the module fuse, inside a rack to the rack fuse, inside a section to the rack and/or section fuse, inside a container to the section and/or container fuse where the PCS DC side is unfused, and a fault fed from the PCS to the container fuse, or the section fuse where there is no container level. Read that column as fault location, not fuse location. The module fuse's stated job is narrower than most people assume: protecting modules from fault currents during transportation, before installation.

The DC-side fuse does not work alone. No single overcurrent protective device covers the range from no-load to maximum prospective short circuit, so the architecture is two devices in series — a resettable DC contactor doing make and break duty, protected upstream by an OCPD covering faults greater than the contactor's breaking capacity at the system voltage.

The guide calls that coordination critical and not trivial, and states three requirements: fuse minimum breaking capacity at or below the contactor's maximum current, the fuse opening fast enough to protect the contactor, and the contactor able to open every overcurrent below the fuse MBC. Drawn out, it is the fuse time-current curve overlaid on the contactor's failure limits — normal thermal limit, contact welding, arcing damage from repulsion.

Typical values and standards

The parts you cite: IEC 60269-1 for general requirements, Edition 5.0 of August 2024; IEC 60269-4 for semiconductor fuse-links, Edition 5.0:2009, superseded by Edition 6.0:2024; IEC 60269-6 for solar photovoltaic energy systems, covering fuse-links for PV strings and arrays up to 1500 V DC; and IEC 60269-7:2021 for batteries and battery systems, also up to 1500 V DC.

Part 6 is the closest precedent worth borrowing from: its tables are headed "gPV" fuse-links, it defines a PV string fuse-link as one giving short-circuit and overload protection in a string, and requires the fuse's rated DC voltage to exceed the maximum open-circuit voltage of the string. Part 7 clause 5.7.1 defines the battery classes — gBat full-range DC, aBat partial range — and clause 5.7.2 sets a minimum rated breaking capacity of 30 kA, higher values permitted.

Two ratings bound the usable window, and in storage the lower one is the one that gets missed. The NEMA Fuse Section guide defines interrupting rating as the maximum prospective current a fuse is tested to open safely at a specific DC voltage and time constant, and minimum breaking capacity as the minimum at those same conditions; the span between them is the range of currents the fuse can safely open.

It notes that MBC is commonly overlooked, and that it matters here because batteries are current-limited sources — a battery rack typically supplies a fault current of 1 to 12 kA, while a combined system can reach 250 kA or more once racks and sections are paralleled. Battery circuit time constants are short: under 10 ms, against 40 to 100 ms for DC traction.

On a shipping datasheet, Mersen's NH gBat range for 1000–1500 V DC advertises full-range breaking capacity, a minimum breaking capacity of 2·In and an interrupting rating that is size- and time-constant-dependent — 50 kA at L/R = 1 ms for the NH1XL and NH2XL sizes, 150 kA at 3 ms for NH3L up to 400 A and 200 kA at 3 ms for NH3L from 450 to 500 A — declared to IEC 60269-7 and UL 248-13; Littelfuse quotes DC breaking capacity up to 250 kA at 1500 V DC together with a low minimum breaking capacity for lower fault-current levels.

NEMA reports that gBat fuse-links must clear twice rated current in the Part 7 Table 104 breaking-capacity test — as of mid-2026 that figure could not be checked against the standard's own text, so treat it as reported rather than quoted. For sizing, Mersen states a minimum of 1.25 × Irms; for selectivity between two gG fuses in series, a rating ratio of about 1.6.

How it shows up in specs, studies and contracts

The curve on a fuse datasheet is a pre-arcing time-current characteristic: pre-arcing time in seconds against the RMS value of the pre-arcing or melting current, with a tolerance band — ±10% on the Mersen gBat sheet. Because it is plotted against an RMS pre-arc current, the same curve serves AC and DC.

IEC 60269-4 Ed. 5.0:2009 states the DC caveat: the pre-arcing characteristic is of particular significance for times exceeding 15τ and is identical to the AC characteristic in that zone, 15τ being chosen to avoid the effects that different rates of rise of current have at shorter times. Below 15τ the standard says the information is conveniently expressed as a pre-arcing I²t characteristic. Above 15τ on DC — or 0.1 s on AC — the arcing period is negligible and operating time equals the maximum pre-arcing time.

I²t is where the melting-versus-clearing split becomes contractual. A gBat datasheet publishes pre-arcing I²t and total clearing I²t separately, and qualifies the clearing value by rated voltage and a stated time constant: Mersen's NH gBat quotes it at Un with L/R = 2 ms. That qualifier is not fine print.

DC clearing I²t changes with voltage, fault current and the circuit time constant, and because there is a different let-through curve for every value of L/R, manufacturers publish none. Mersen's worked DC example makes the split concrete — 21.6 ms pre-arcing, an estimated 25 to 30 ms of arcing, 50 ms total clearing, and pre-arcing I²t of 295 700 A²s. The energy liberated in 1 ohm of the protected circuit equals the numerical value of the operating I²t in A²s.

A DC fuse spec reading only "1500 V, 400 A" is unusable. Demand the rated DC voltage with its tested time constant, the interrupting rating and minimum breaking capacity at those same conditions, both I²t values, and the contactor coordination overlay. IEC 60269-4 clause 5.2 is the reason to be pedantic: a fuse-link shall have an AC voltage rating or a DC voltage rating or a VSI voltage rating, and may have one or more — an AC rating implies nothing about DC.

In US jurisdictions NEC 2020 Article 706.31 adds: a DC rating, adequate interrupting rating, sizing at not less than 125% of maximum current, and a listed, labelled current-limiting device adjacent to the ESS for each DC output circuit unless the listed ESS already provides it. NEC 706 was revised in the 2023 cycle — cite the edition.

Common pitfalls

The expensive one is specifying a partial-range fuse as the only protective device. An aM or aBat fuse-link is defined to break currents between k2·In on its operating characteristic and its rated breaking capacity — below that band it is not tested to clear anything. Mersen's training material is blunt about the motor case: an aM fuse must be associated with other protective devices because it must not operate for times above 60 seconds.

In a BESS the fuse-plus-DC-contactor pairing rests on a related but distinct rationale — no single device spans no-load to maximum prospective fault — and manufacturer literature points to full-range gBat at rack level (Mersen sells its NH gBat range for exactly that duty), though secondary summaries of Eaton Bussmann's battery-storage application guide describe aBat rack fuses instead; that guide could not be obtained for this entry, so treat the contrast as reported rather than verified, and confirm the class against the fuse actually offered.

If a bid sheet shows an "a" class with no companion switching device named, that is a finding, not a detail.

Second: carrying numbers between parts of IEC 60269. Part 4 clause 5.7.2 recommends a breaking capacity of at least 50 kA for AC and 8 kA for DC — that is a semiconductor-fuse recommendation, and quoting it on a battery job understates the requirement, because IEC 60269-7 clause 5.7.2 sets a 30 kA minimum.

The same clause is worth reading for how ratings are earned: the AC rating comes from type tests in a circuit containing only linear impedance with a constant sinusoidal applied voltage, the DC rating from a circuit containing only linear inductance and resistance at mean applied voltage. The standard itself notes that non-linear impedances and unidirectional voltage components in service can shift breaking severity either way.

Third: treating minimum breaking capacity as a footnote. A rack fuse wants a low MBC, in the range of 2 to 3 × In, precisely so it coordinates with and protects the DC contactor, while for a section fuse the interrupting rating is the critical parameter — an internal rack fault can expose the rack fuse to cascading contributions from adjacent racks. Two smaller traps close the list.

Clearing I²t is not one number on DC, so a single figure quoted without voltage, fault current and time constant is not comparable between vendors. And the vocabulary drifted: "utilization category" in IEC 60269-1 Ed. 3.1 clause 2.2.10 became "utilization class" at clause 4.9 in Ed. 5.0:2024, while IEC 60269-7:2021 still says category. Same concept, two words.