Broken rotor bar detection is the process of identifying cracked, fractured, or fully severed conductor bars in the squirrel cage of an induction motor before they cause thermal runaway, cage disintegration, or consequential damage to the stator and bearings. Rotor bars carry the induced currents that produce torque, and once one bar fails, the current it should have carried redistributes into neighbouring bars, accelerating their failure. Left undetected, a single broken bar typically progresses to multiple broken bars and rotor cage failure within months, so early detection through electrical, magnetic, or vibration-based methods is a core reliability discipline for any plant running large induction motors.
Squirrel cage rotors are built from aluminium (die-cast) or copper bars joined to end rings, sitting inside laminated steel slots. Failure is almost always driven by a combination of thermal, mechanical, and electromagnetic stress rather than a single cause.
Broken bars rarely announce themselves through an obvious single symptom. Watch for:
MCSA is the primary non-intrusive method for broken rotor bar detection. It requires only a current clamp or permanently installed CT on one stator phase, high-resolution FFT analysis, and the motor running under load, no disassembly needed. This makes it a natural fit for route-based condition monitoring programmes, where current signatures are trended alongside vibration data.
A broken bar interrupts current flow at that location, creating an asymmetry in the rotor's magnetomotive force. This asymmetry produces a backward-rotating magnetic field component that induces a modulation on the stator supply current at twice slip frequency. The result is a pair of sidebands around the fundamental line frequency component.
The characteristic broken rotor bar sidebands appear at:
f_b = (1 ± 2s) × f
Where f is the supply frequency (50 or 60 Hz) and s is the per-unit slip, defined as:
s = (n_s − n_r) / n_s
with n_s the synchronous speed in rpm (n_s = 120f / p, where p is the number of poles) and n_r the actual rotor speed in rpm.
Worked example: a 4-pole, 50 Hz motor has synchronous speed n_s = 120 × 50 / 4 = 1500 rpm. If the rotor runs at 1470 rpm under load, slip s = (1500 − 1470) / 1500 = 0.02 (2%). The sidebands appear at:
Both sit close to the 50 Hz fundamental, so the spectrum must be captured with fine frequency resolution (0.01 to 0.05 Hz bins, meaning long sample windows of 30 to 100+ seconds) and a stable, steady-state load to resolve them cleanly. Load fluctuation smears slip and blurs the sidebands, which is the single most common cause of a missed or ambiguous reading in the field.
Severity is judged from the sideband amplitude relative to the fundamental component, expressed in dB. This is a widely used industry guideline rather than a single universal standard, and thresholds vary somewhat by reference and motor design.
| Sideband level relative to fundamental | Typical interpretation | Recommended action |
|---|---|---|
| Below −54 dB (below noise floor) | No detectable defect | Continue normal monitoring interval |
| −54 to −45 dB | Early-stage indication, possibly a single hairline crack or high-resistance joint | Trend monthly; verify at consistent load |
| −45 to −35 dB | One or two broken bars, or a broken end-ring segment | Increase monitoring frequency; plan inspection at next outage |
| −35 to −30 dB | Multiple broken bars, cage degradation advancing | Schedule corrective maintenance; avoid repeated hard starts |
| Above −30 dB | Severe cage damage, risk of bar fragments damaging windings | Plan removal from service; rotor repair or replacement |
Note that the number of broken bars does not scale linearly with sideband amplitude: cage geometry, bar location relative to the flux pattern, and rotor-to-stator eccentricity all influence the reading, so MCSA severity bands should be treated as a maintenance planning aid rather than an exact bar count.
MCSA is rarely used in isolation for a critical asset. Cross-checking with other techniques reduces false positives, since bearing wear, gear mesh issues, or supply voltage unbalance can produce current or vibration components that mimic broken-bar signatures.
Because broken-bar signatures develop gradually and depend on consistent load and slip conditions, the real value comes from trending, not one-off spot checks. Recording current spectra at the same load point on a fixed interval, logging them against a maintenance history, and automatically opening a work order when a sideband crosses a defined dB threshold turns a manual diagnostic task into a repeatable early-warning system. This is where condition-based maintenance workflows tie directly into a CMMS: an OEE and condition monitoring platform can flag the trend breach, attach the spectrum to the asset record, and generate the inspection work order automatically, closing the loop between the electrical signature and the maintenance action before a bar fragment reaches the stator winding. Fabrico supports this kind of condition-triggered workflow as part of its broader OEE and maintenance tracking.
Related failure modes worth monitoring alongside broken rotor bars include motor slip in induction machines and partial discharge testing for stator winding insulation, since rotor and stator degradation often accelerate each other once one fault is present. To see how automated condition triggers and work order generation work in practice, Book a Fabrico demo.
Yes, in the short term. A single broken bar in a large motor often continues operating with only slightly reduced efficiency and increased vibration, but the remaining bars now carry higher current density, which accelerates fatigue in adjacent bars. Continued operation should be based on a documented severity assessment and load reduction where possible, not left indefinite.
It is more difficult. VFD output contains switching harmonics that can mask or mimic the (1±2s)f sidebands, and slip calculation requires knowing the actual commanded frequency at each moment. Specialised MCSA tools with drive-aware filtering are needed, and results are generally less reliable than on line-fed motors.
Voltage or load unbalance, rotor eccentricity, gearbox or coupling misalignment, and even normal manufacturing asymmetry in the rotor cage can all produce sidebands near the broken-bar frequencies. This is why cross-referencing MCSA with flux or vibration data before condemning a rotor is standard practice.
Higher pole-count motors run at lower synchronous speed and often operate at higher per-unit slip, which pushes the sidebands further from the fundamental and makes them easier to resolve. Low-slip, lightly loaded 2-pole motors are the hardest case, since the sidebands can sit within a fraction of a hertz of the fundamental.
Anmeldung
Blog