---
title: "Busbar sizing: current rating, heat and short-circuit forces"
description: "Why amps per mm² is not a constant, what paralleling really buys, and the mechanical force a fault puts on the supports — with the arithmetic worked."
date: "2026-03-28"
author: "Divakar B"
source: "https://energycalchq.com/blog/busbar-sizing-rating-and-forces"
license: "© 2026 Divakar B. Quote with attribution to https://energycalchq.com/blog/busbar-sizing-rating-and-forces"
---

A busbar has to do two unrelated things. It has to carry the continuous current
without getting too hot, and it has to survive a fault without being torn off
its supports. The first is a thermal problem, the second is mechanical, and they
are sized by completely different arithmetic.

Panels usually get the first roughly right using a rule of thumb, and never
check the second at all.

## Amps per mm² is not a constant

Ask around a panel shop and you will be told copper carries 1.6 A/mm². Apply
that to a 12 × 3 bar and you get 58 A, when the real figure is nearer 110. Apply
it to a 120 × 10 and you get 1,920 A, which is optimistic by a wide margin.

The rule is only true near the size the person quoting it usually works with.

The physics is straightforward. A bar generates heat throughout its
**cross-section** and sheds it from its **surface**. Double the area and the
surface does not double, so the current each square millimetre can sustain has
to fall.

![Curve of copper busbar current density falling from 3.15 to 1.38 A/mm² as bar area increases](/blog/busbar-current-density.svg "The rule of thumb is a horizontal line through a curve. It crosses the truth at about 600 mm² and is wrong everywhere else.")

Fitting a power law to published manufacturer tables gives a usable model:

```
I ≈ C × A^0.765        A in mm²
C = 7.3 copper, 4.6 aluminium
(bare bar, still air, 30 K rise over 40 °C ambient)
```

That tracks real tables within a few per cent from 12 × 3 up to 120 × 10, which
is close enough to choose a bar and quote it. The [busbar sizing
calculator](/tools/busbar-sizing) uses it directly.

It is not a substitute for a type test. An assembly to IS/IEC 61439 is rated by
**measured temperature rise on the actual panel**, and the enclosure, the joint
quality and the cable entries all change the answer. Calculate to choose the
bar; test to declare a rating.

## Paralleling has diminishing returns

Bars in parallel do not add up:

| Bars per phase | Capacity vs one bar |
|---|---|
| 2 | 1.8× |
| 3 | 2.5× |
| 4 | 3.1× |

Two effects work against you. The inner faces of bars sitting 5–10 mm apart
radiate into each other rather than into the air, and at 50 Hz the proximity
effect pushes current towards the outer faces so the metal in the middle carries
less.

By the fourth bar you are paying for copper that does very little. Past three or
four bars per phase, a sandwich arrangement or a type-tested busbar trunking
system is the better buy.

This is also why bar *shape* matters more than bar *area* at large sizes. Skin
effect confines current towards the surface at 50 Hz, so a tall thin section
outperforms a square one of the same area — the tall bar has more surface for
the same metal, both for carrying current and for shedding heat. It is the
reason busbars are flat strips rather than square rods, and the reason very
large systems move to tubular or channel sections rather than simply stacking
more flat bars.

## Derate for the panel, not the room

The ambient that matters is the air **inside the enclosure**, and on a loaded
panel in an Indian plant room that runs 10–15 °C above the room temperature.

| Condition | Factor |
|---|---|
| 40 °C internal ambient | 1.00 |
| 45 °C | 0.95 |
| 50 °C | 0.90 |
| 55 °C | 0.84 |
| Ventilated enclosure | 0.90 |
| Sealed IP54+ enclosure | 0.80 |

A 50 °C internal ambient in a sealed enclosure is 0.90 × 0.80 = 0.72. Nearly a
third of the bar rating gone, and none of it visible on a drawing that just says
"1600 A busbar".

## Now the mechanical half

Two parallel conductors carrying current in opposite directions push each other
apart. The force goes with the **square** of the current:

```
F = 2 × 10⁻⁷ × I²peak / d      newtons per metre
```

Use the peak asymmetrical current, roughly 2.5 times the symmetrical rms value
for LV systems. Worked through for a 25 kA board with bars 75 mm apart:

```
Peak = 25,000 × 2.5 = 62,500 A
F    = 2 × 10⁻⁷ × 62,500² / 0.075
     = 10,417 N/m
```

**Over a tonne per metre**, for a few milliseconds, trying to tear the bars off
their insulators. That is what busbar supports are for, and it is why their
spacing tightens as fault level rises.

Two relationships worth carrying in your head:

- Halve the phase spacing and you **double** the force.
- Double the fault current and you **quadruple** it.

Which means a panel that was fine on a 10 kA supply is not automatically fine
after the transformer is uprated. Fault level scales with transformer kVA — check
it with the [transformer sizing calculator](/tools/transformer-sizing) — and the
bracing has to be rechecked along with the breaking capacity.

## Two ratings, two failure modes

It is worth being precise about the two short-circuit numbers, because they get
conflated:

| Rating | Symbol | What it means |
|---|---|---|
| Short-time withstand | Icw | The rms current the assembly carries for 1 s (or 3 s) without damage — a **thermal** limit |
| Peak withstand | Ipk | The instantaneous peak it survives without mechanical failure — a **mechanical** limit |

A busbar can have plenty of one and not enough of the other. Icw is about the
copper getting too hot; Ipk is about the supports letting go. Both must exceed
what your system can deliver.

## Joints are what actually fail

In service, panels rarely fail because the bar was a size small. They fail at the
joints, and the failure mode is always the same: a joint with poor contact heats,
which oxidises the surface, which worsens the contact, which makes it hotter.

- **Overlap by at least the bar width**, with a minimum of two bolts.
- **Torque to specification** with a torque wrench. Under-tightened joints run
  hot; over-tightened ones creep and relax.
- **Belleville washers on aluminium**, always. Aluminium creeps under sustained
  load and a flat-washer joint loses its clamping force within a year.
- **Clean and compound aluminium faces immediately** before bolting. Aluminium
  oxide forms within minutes and it is an insulator.
- **Bimetallic connectors** where aluminium meets copper. Direct contact in a
  damp panel is a galvanic cell.
- **Thermal-image the panel at full load** after commissioning, and again
  annually. Every joint should be within a few degrees of its bar. The one that
  is not is the one that will fail.

## Clearances are part of the busbar design

The bars themselves are only half the geometry. Air is the insulation between
them, and how much you need depends on the voltage and on how dirty the air is
going to get.

For a 415 V system, common working minimums inside an assembly are 20 mm
phase-to-phase and 20 mm phase-to-earth in clean conditions, opening up as the
pollution degree rises. IS/IEC 61439 sets them properly against pollution
degree and impulse withstand, and a type-tested assembly has them verified
rather than assumed.

Two practical points that catch people:

- **Creepage is not clearance.** Clearance is the shortest path through air;
  creepage is the path across an insulating surface. Dust and condensation
  track across surfaces, so creepage distances are longer, and a support
  insulator with a ribbed profile is buying creepage rather than looking
  decorative.
- **Sleeving the bars does not license shorter spacing** unless the assembly
  was tested that way. Heat-shrink sleeving is a barrier against accidental
  contact and a phase-identification aid; treating it as insulation that
  permits closer bars is a common and untested assumption.

Remember also that reducing phase spacing to fit a narrower chamber doubles the
short-circuit force for every halving of the gap. The clearance decision and
the bracing decision are the same decision.

## Sizing the neutral and earth bars

The phase bars get the attention and the other two get a rule of thumb, which
is usually fine and occasionally badly wrong.

**The neutral** is traditionally half the phase cross-section on a balanced
three-phase board, and that holds where the load really is balanced and linear.
It stops holding where triplen harmonics are present — third-harmonic currents
from single-phase electronics and LED drivers are in phase across all three
lines and add in the neutral instead of cancelling. On an office or retail
board, a full-size neutral is the safer specification, and where third harmonic
exceeds about a third of the line current the neutral is the conductor that
sizes the chamber.

**The earth bar** is sized on fault energy, not on load: `S ≥ √(I²t) / k`, the
same adiabatic relationship that sizes a protective conductor. It carries
nothing at all until something fails, and then it carries everything for as
long as the breaker takes to clear — which is why a slower upstream device
means a larger earth bar, not just a larger cable.

## Specifying a busbar chamber

1. Continuous current — usually the **incomer rating**, not the connected load.
2. Bar material and arrangement, derated for the internal ambient and the
   enclosure.
3. Icw and Ipk at or above the system fault level, from the transformer
   impedance.
4. Support spacing checked against the calculated force.
5. Phase spacing and clearances stated on the drawing, because they change the
   force.
6. Joint torque specified, and re-torque scheduled at first maintenance.

Run the numbers with the [busbar sizing calculator](/tools/busbar-sizing) — it
selects a standard bar section for your current and conditions, and reports the
force per metre and an indicative support spacing from your fault level.