Electrical Monitoring: How to Spot Faults Early

The Failure You Can See Coming

Ask a maintenance team how they catch problems early and most will say vibration. That answer is right, and it is incomplete.

Vibration analysis is excellent at what it does. Bearing wear, misalignment, imbalance, looseness, and gear defects all produce mechanical motion you can measure and trend. But a large share of what kills industrial motors is not mechanical in origin. It is electrical. A degrading stator winding, a supply unbalance quietly cooking insulation, a cracked rotor bar, a harmonic environment overheating a motor that looks fine on every mechanical measurement. Those faults live in the current and voltage waveforms, and some of them never produce a mechanical symptom until the moment they produce a failure.

That is the case for electrical monitoring. Not as a replacement for vibration, but as the other half of the picture. This guide covers what to measure, what the numbers should look like, which signatures indicate which fault, and how to build a program that catches problems while they are still cheap.

The Six Parameters That Carry the Signal

Electrical monitoring means continuous or periodic measurement of the electrical parameters feeding and flowing through an asset. Six of them do most of the diagnostic work.

Current. The workhorse measurement. Current reveals load changes, phase loss, and mechanical problems reflected back through the motor. A pump fighting a partially blocked strainer shows up as a current change before anything else.

Voltage. Tells you about supply quality and asymmetry between phases. Voltage problems often originate upstream of the asset, which means the asset is the victim rather than the cause.

Power factor. Indicates how much of the delivered power is doing useful work. Shifts in power factor point to changes in load character or motor condition.

Total harmonic distortion. Captures waveform distortion introduced by variable frequency drives, rectifiers, and other non-linear loads. Harmonics create heating that no mechanical measurement will ever see.

Insulation resistance. Measures the integrity of winding and cable insulation. This is the parameter that tells you how much life the motor has left in it.

Phase balance. Assesses symmetry across the three supply phases. Small asymmetries here produce disproportionate consequences, which is worth its own section below.

What the Numbers Should Look Like

Generic thresholds are a starting point, not a program. Use them to catch obvious problems while you build asset-specific baselines. Reasonable alert points:

• Current imbalance above 5 percent. Investigate.
• Voltage imbalance above 2 percent. Investigate, and treat 1 percent as the real target. In the same tip sheet above, the Department of Energy recommends voltage unbalance at motor terminals not exceed 1 percent, notes that anything above that requires derating the motor per NEMA MG-1, and warns that operating above 1 percent unbalance "will void most manufacturers' warranties."
• Total harmonic distortion above 8 percent on motor circuits.
• Insulation resistance below 1 megohm per kilovolt of rated voltage. This one is a floor, not a trend line, and crossing it means the asset needs attention now.

The more useful discipline is comparing each asset against itself. A motor that has run at 3 percent current imbalance for four years is telling you something different than a motor that moved from 1 percent to 3 percent last month. The absolute number matters less than the direction and the rate.

The Fault That Costs You Twice

Voltage unbalance deserves a section of its own, because it is the clearest example of why early detection matters and because most plants underestimate it badly.

Start with how it amplifies. Current unbalance runs 6 to 10 times as large as the voltage unbalance that causes it. In one Department of Energy tested example (in the same tip sheet above), a 2.5 percent voltage unbalance produced a 27.7 percent current unbalance. A supply problem you might dismiss as minor arrives at the motor windings as a serious one.

Then look at what that does to temperature. The Department of Energy gives the relationship as:

Total Temperature Rise = Balanced Temperature Rise x (1 + 2 x (% Voltage Unbalance)² / 100)

Run a motor with an 80 degree Celsius balanced temperature rise at 2 percent voltage unbalance and the rise becomes 86.4 degrees. That looks survivable until you add the rule that governs insulation: winding insulation life is cut in half for every 10 degree Celsius increase in operating temperature. Unbalance does not break the motor today. It quietly spends the motor's remaining years.

And it costs you on the meter while it does it. In the same Department of Energy example, a 100 horsepower motor at full load dropped from 94.4 percent efficiency to 93.0 percent under 2.5 percent voltage unbalance. You pay for the fault in electricity every hour it runs, then pay again when the winding fails early.

This is the whole argument for electrical monitoring in one fault. A condition that produces no noise, no vibration, and no obvious symptom is measurably shortening asset life and inflating your energy bill, and the only way to see it is to measure the electrical side.

Reading the Current Waveform

Electrical signature analysis (ESA) treats the current waveform as a diagnostic record. Faults inside the motor modulate the current in patterns specific to what is wrong. The main signatures:

Broken or cracked rotor bars produce sidebands at the supply frequency plus and minus the slip frequency. This is one of the cleanest diagnoses in the discipline and one that vibration frequently misses under light load.

Air gap eccentricity, both static and dynamic, creates sidebands tied to the supply frequency and the rotor slot harmonics.

Bearing defects appear as sidebands offset from the supply frequency by the bearing defect frequencies, meaning outer race, inner race, ball spin, and train frequencies. Useful when a sensor cannot be mounted where vibration would need it.

Stator winding faults and insulation degradation show as changes in harmonic content, particularly in the third harmonic and other odd harmonics.

Load-side problems such as misalignment, gear wear, and cavitation modulate current at frequencies matching the load event rate. The motor becomes a sensor for everything downstream of it.

Severity is read from the amplitude of these sidebands relative to the fundamental, measured in decibels.

One important operating condition: accurate analysis generally requires the motor to be running at 40 percent load or higher. Below that, fault signatures weaken and sink into the noise floor. A test run on an unloaded motor can return a clean result on a motor that is not clean, which is a failure mode of the method rather than of the equipment.

Where This Pays Off Across Sectors

The parameters are universal. What varies is which assets carry the consequence.

• Food and beverage. Refrigeration compressors and CIP pumps, where a winding failure risks product alongside the repair, and where continuous duty means a small efficiency loss compounds all year.
• Metals and heavy manufacturing. Large VFD-driven drives, where harmonics and supply disturbances are constant and where drive stress precedes motor damage.
• Chemical and process. Pumps, agitators, and blowers, often in classified areas where an electrical fault is a safety exposure and not only a maintenance one.
• Mining and aggregates. Conveyor and crusher drives, where access is hard and current analysis can diagnose from the MCC rather than at the machine.
• Pharmaceutical and life sciences. Chillers and utility systems tied to batch validity, where insulation trending protects far more value than the motor is worth.

Building a Program That Actually Catches Things

Baseline every critical motor while it is healthy. You cannot trend what you never measured. A baseline taken on a known-good asset at known load is the most valuable data point in the whole program, and it is only available before the problem starts.

Set thresholds per asset, not per plant. Motor size, duty cycle, drive type, and supply environment all shift what normal looks like. Generic limits generate false alarms on some assets and stay silent on others.

Trend, do not spot-check. A single reading tells you where an asset is. A trend tells you where it is going and how fast, which is the only information that lets you plan.

Test at representative load. Schedule electrical readings during normal production, above 40 percent load, and record the load with the reading so comparisons are valid.

Pair electrical with mechanical. Vibration and electrical monitoring fail to see different things. Run together, they cover bearings, alignment, and looseness on one side and windings, rotor, supply quality, and insulation on the other. Neither one covers the other's blind spot.

Route alerts into work orders. An electrical monitoring program that produces alarms nobody schedules against delivers nothing. The alert has to become a task, with an owner and a date, or the measurement was academic.

The Mistakes That Hide Faults

Treating unbalance as the utility's problem. Some of it is. Much of it comes from unequal single-phase loading, loose connections, and failing contactors inside your own plant, all of which you can fix.

Testing at low load and trusting the result. Covered above, and worth repeating because it produces false confidence rather than a missed reading.

Watching amps only. Current alone will not distinguish a supply problem from a load problem from a winding problem. The parameters are diagnostic in combination.

Ignoring energy per unit produced. Rising kilowatt-hours per unit of output is one of the earliest and least ambiguous signals of electrical degradation, and it usually already exists in data the plant collects for other reasons.

Where Tractian Fits

Most plants end up with electrical and mechanical monitoring in separate systems, run by separate people, on separate schedules. The faults do not respect that boundary, and neither should the tooling.

Tractian is one of the few platforms that combines dedicated mechanical sensing with electrical monitoring in a single analytics environment, so a rotor bar signature in the current waveform and a bearing trend in the vibration data land in the same asset record, reviewed by the same team.

Energy Trac handles the electrical side. It captures current and voltage per phase, power factor, harmonics, and phase angle continuously, installs non-invasively with no process interruption, and applies electrical signature analysis to detect rotor, stator, winding, and insulation faults. Our engineering team describes the goal plainly: surface electrical fault signatures in the current waveform months before secondary damage appears. It also tracks kilowatt-hours per unit produced, which turns the energy meter into a condition monitoring instrument.

If you are running vibration today and suspect you have a blind spot on the electrical side, we can show you what your critical motors are already telling you.

Ready to see it on your assets? Schedule a demo.