Duty groups: why two cranes of the same capacity differ

Two cranes rated for the same load can need entirely different drives, gearboxes and structures. The rated load says nothing about how often it is lifted, or how heavy the average lift is. Crane engineering therefore classifies a machine by a duty group that combines two independent measures. This lesson is course section 4.1.2.

Two measures

Class of utilisation: the total number of cycles the mechanism must perform in its design life. A maintenance hoist in a pumping station may make a few thousand lifts in twenty years; a ladle crane in a steelworks makes that many in a month.
Load spectrum (state of loading): how the loads are distributed across those cycles. A crane that lifts its rated load on nearly every cycle has a far harsher fatigue history than one whose average lift is a fifth of rated, even with the same number of cycles.

The load spectrum is summarised by the spectrum factor

k_p = \sum_i \frac{n_i}{N}\left(\frac{P_i}{P_\text{max}}\right)^3 \tag{4.3}

where $n_i$ cycles are made at load $P_i$ out of $N$ in total, and $P_\text{max}$ is the rated load. The cube reflects the fatigue of the steel structure and the gearing: a load at half of rated does only one eighth of the damage of a full load. Light lifts fill most of the cycles but contribute almost nothing.

Try it: load spectrum and fatigue

Share of cycles
Share of the damage (cube law)

Lifts at 20 % of rated60 %

Lifts at 40 % of rated20 %

Lifts at 60 % of rated10 %

Lifts at 80 % of rated7 %

Lifts at 100 % of rated3 %

Spectrum factor k_p: 0.11
Spectrum class: L1
Damage from the two heaviest levels: 63 %

k_p = Σ (n_i / N) (P_i / P_max)³

Predict first

In the 'Light (L1)' preset, 60 % of the lifts are at 20 % of rated load. What share of the damage do they cause?

About 60 %About 20 %Under 5 %

Duty groups

Combining the class of utilisation and the load spectrum gives the duty group: M1 to M8 in ISO 4301 and FEM 9.511, Class A to F in the North American CMAA specification. The spectrum itself is classed L1 (light, $k_p \le 0.125$ ), L2 ( $\le 0.25$ ), L3 ( $\le 0.5$ ) and L4 (very heavy, up to 1).

Group	CMAA class	Typical service
M3	B	maintenance and assembly hoists; infrequent, light average load
M5	C/D	general workshop and warehouse cranes; moderate regular duty
M6	D/E	foundry, scrapyard and process cranes; heavy, frequent duty
M7–M8	F	ladle, grab and magnet cranes in continuous production; near-rated load on almost every cycle

These are design inputs, not labels: the duty group fixes the fatigue life of the structure, the service factor on the gearbox, the minimum rope safety factor and the thermal basis of the motor. It does not change the rated capacity.

What it means for the drive

Thermal rating. A crane motor is rated on a cyclic duration factor (CDF, Einschaltdauer): the energised share of each cycle, quoted with a cycle length, typically 40 % CDF at 10 min. This is duty type S3 to S5 from Chapter 1, not continuous S1; a drive chosen on its continuous rating alone will be undersized for a high-group crane.
Starts per hour. A group M7 hoist may exceed 300 starts an hour. That rules out direct-on-line switching altogether: an inverter is mandatory on thermal grounds, before any control benefit.

FoundationStart here if this is new to you

Think of two cars that can both carry 500 kg. One is a family car that carries a full load once a year; the other is a delivery van loaded to the limit forty times a day. On paper they have the same capacity; in practice the van needs stronger springs, brakes and engine cooling. A crane's duty group is the difference between the family car and the van.

ExplorerGo deeper: derivations and open questions

Design life. A mechanism is designed for a total of $N$ cycles. Show that a crane with $k_p = 0.125$ can make eight times as many cycles as one with $k_p = 1$ for the same fatigue damage. What does that say about using a light-duty crane for heavy production?

Equivalent load. The cube root $\sqrt[3]{k_p}\,P_\text{max}$ is the constant load that would do the same damage. Compute it for the three presets. How does it compare with the arithmetic mean load?