Troubleshooting
Diagnosing 200 A currents on equipment grounding conductors
Short answer
Hundreds of amperes on equipment grounding conductors between a substation and its MCCs, with like phases grouped in cable tray, is magnetically induced circulating current, not a ground fault. Clamp all circuit conductors of one feeder: a residual near zero proves induction, and the fix is regrouping each parallel set A-B-C.
On this page
The 200 A you are reading on equipment grounding conductors between the substation and the MCCs is magnetically induced circulating current, not a ground fault. With four parallel conductors per phase and all the A cables laid in one plane, then all the B, then all the C, the three phase fields cannot cancel, and the 300 ft tray run behaves like the primary of a loosely coupled transformer. Every closed loop made by the grounding conductors, the tray, the ground buses and the building steel is a secondary winding, and the induced voltage drives current around all of them. One measurement separates this from a fault: clamp all the phase conductors of one feeder and read the residual. The fix is regrouping each parallel set A-B-C with the feeder locked out — never more bonding, and never an opened ground.
Why like-phase grouping in tray induces ground current #
Field cancellation is a geometry result. When the A, B and C conductors of one circuit sit against each other, in trefoil or side by side, their fields nearly cancel a short distance away. This installation spread each feeder across the tray instead: four A conductors together, then four B, then four C, so at any point above the run the fields from the three phases arrive with different strengths and add instead of cancelling. The single 4/0 grounding conductor lay on top of the groups, which gives it unequal coupling to the three phases and a net induced voltage along its length.
The circuit closes through everything bonded at both ends: the grounding conductors between substation and MCCs, the tray sections, the MCC ground buses, the substation ground bus, the building steel, and every jumper between them. Induced voltage appears around each of those loops, and current circulates the whole time the load is on. Magnitude scales with total load current, the width of the phase groups, the distance from the groups to each grounding path, the route length, the number of parallel loops and the impedance of each bond — which is why 5,000 A of load on 300 ft of misgrouped conductors produces hundreds of amperes, and why an added bond acquires its own share instead of relieving the others.
Separating induced current from a ground fault #
Run these checks before anyone moves a conductor, with the energized readings taken under your site's electrical safety program:
- Residual test on one feeder. Clamp all phase conductors of one feeder in a single pass — twelve single conductors here — plus the neutral if the circuit has one, excluding all equipment grounding conductors. A flexible core-balance probe or a large-jaw clamp meter rated for the circuit does it; zero the instrument first. A residual of a few amperes against the feeder's full load means no current is leaving the circuit, so the grounding-conductor current is induced. A residual of tens of amperes or more means leakage or a fault — stop and diagnose that feeder.
- Switch one feeder at a time under an approved switching plan and record which grounding currents fall, persist or transfer to another path. Induced current tracks the combined field of all feeders; leakage tracks its own load and nothing else.
- Inspect the neutral system with the equipment de-energized. A neutral-to-ground bond downstream of the source, or a load returning current through the grounding path, produces continuous ground current and must be corrected regardless of what else you find.
- Read phase-to-ground voltages. Unbalanced phase-to-ground voltage moves the investigation to fault diagnostics first. Balanced readings do not clear the grounding conductors; they are incomplete evidence either way.
- Compare the motor grounds. Almost the same current on many unrelated motors, 15 A each here, is a shared mechanism, not dozens of simultaneous motor insulation failures.
Worked check: 900 A of ground current against 5,000 A of load #
The inputs, from this installation's records and readings:
- 480 V double-ended substation, two 3000 kVA 12.5 kV–480 V transformers, six MCCs on 1600 A buses carrying about 1000 A of load each.
- Six feeders of four single-conductor 500 kcmil per phase, 300 ft in cable tray, like phases grouped in one plane, one single-conductor 4/0 equipment grounding conductor laid on top of the groups.
- Measured: about 5,000 A summed across all phase conductors, up to 200 A on some grounding conductors between the substation and the MCCs, about 900 A summed across the grounding paths, about 15 A on each motor grounding conductor, and 10 A on an added 4/0 jumper from an MCC ground bus to building steel with no change on the original paths.
The rule in words: fault or leakage current leaves the circuit conductors and shows up as residual; induced current stays inside closed bonding loops and grows with total load current. So compare the two sums, then run the residual test on one feeder.
The arithmetic:
- Average feeder load: 5,000 A ÷ 6 feeders ≈ 833 A.
- Circulating share: 900 A ÷ 5,000 A = 0.18, so about 18% of the installation's load current circulates on grounding paths — a comparison for this installation, not a limit of any kind.
- Residual check on one feeder: twelve phase conductors in one clamp, expecting a reading within the instrument's accuracy of zero while the grounding conductor alone reads 200 A.
The decision: with a near-zero residual, balanced phase-to-ground voltages, near-identical motor ground currents and a new bond that picks up 10 A without reducing anything, commit to induced circulating current from conductor grouping, and spend the outage regrouping conductors rather than adding grounding. If the residual had read tens of amperes or more, the same numbers would point to leakage or a fault on that feeder, and diagnostics would come before any tray work. These sums are estimates from clamp readings; confirm the repair scope against the governing code and the equipment documentation.
Fixes that cannot work #
- More bonding. The 4/0 jumper from the MCC ground bus to building steel carried 10 A and left the 200 A paths untouched. Every added bond is another closed loop sharing the same induced voltage; none of them removes it.
- Adding neutrals the loads do not need. A neutral fixes current that returns through the grounding path — correct an actual neutral-to-ground violation — but it does nothing to a magnetic coupling problem.
- Opening or lifting a grounding conductor. That removes the fault-clearing path the whole scheme depends on, and the current largely returns through the tray and steel anyway.
- Accepting the condition because voltages look balanced. Balanced phase-to-ground voltage never proved grounding conductors clean; it only failed to prove the opposite.
The repair: regroup each parallel set #
This work is done de-energized: open the feeder, lock and tag it, and test for absence of voltage on every conductor you will touch, following NFPA 70E, 2024 edition. Nothing in this repair happens on live 480 V conductors.
- Before the shutdown, photograph each tray run and build a conductor schedule: which conductors exist, their positions, crossings, vertical layers, and where every grounding conductor and bond sits. Record the exact number of grounding conductors per feeder — the count changes both the readings and the scope.
- Confirm the phase identification of all twelve conductors of each feeder before cutting a single tie.
- Establish the electrically safe work condition on the feeder: open, lock, tag, test for absence of voltage, and verify the tester on a known source.
- Cut the ties and rebuild each feeder as four bound groups of one A, one B and one C conductor lying adjacent, distributed across the tray. Keep each equipment grounding conductor with its group, not parked above one phase set.
- Re-tie each group securely, verify tray section bonding, and restore the covers.
Two code points govern the result. NFPA 70, the National Electrical Code, 2023 edition, Article 392 requires single conductors installed in cable tray to be bound in circuit groups, which the regrouping restores.
NFPA 70, 2023 edition, 250.122(F) requires each parallel run to carry its own equipment grounding conductor sized from Table 250.122 for the feeder's overcurrent device; the 2017 edition is what first separated the rules for multiconductor cables from those for individual parallel conductors. Copper 4/0 is the Table 250.122 size for a device above 1200 A through 1600 A, so with 1600 A feeder protection each of the four conductor groups needs one that size, and a single remote conductor for the whole set raises the ground-fault path's inductive reactance. Section 250.6 treats current on grounding paths during normal operation as objectionable and directs you to correct the cause without breaking the grounding connection. Confirm these interpretations with the local authority having jurisdiction and the equipment documentation.
Re-measure at the same load, then close it out #
Repeat every measurement at the same load state with the same instruments: each feeder's grounding conductors, the tray bonds, the transformer neutral connections to the ground bus, and the motor grounds. The 200 A paths should collapse to a small residual, and the 15 A motor grounds should drop with the field that induced them. Any path still carrying a significant share of load current after regrouping points to a loop you have not addressed — commonly a grounding conductor still remote from its phases, or tray sections acting as a parallel conductor. Keep the before-and-after records together; they are the proof that the mechanism was induction, and the baseline if a reading ever climbs again. A motor whose ground current persists after the fix gets its own insulation-resistance test and a check for neutral-to-ground bonds or filter current.
Questions asked
Is induced current on grounding conductors actually harmful, or just a stray reading?
It dissipates real power and hides real faults. At 200 A through 300 ft of 4/0 copper, resistance is about 0.0608 Ω per 1000 ft times 0.3, or 0.018 Ω (NEC Chapter 9, Table 8, DC at 20 °C), so losses are roughly 200 A squared times 0.018 Ω, about 720 W of heat in that one conductor, before AC impedance and temperature raise it. Ground-fault relays that measure residual current can also respond to it. Fix the geometry rather than monitor it.
Our added 4/0 bonding jumper read 10 A and changed nothing elsewhere. What does that prove?
That result is the signature of circulation. Current driven around closed loops divides among every parallel path in proportion to impedance, so a new 4/0 bond simply picks up its share, 10 A here, while the 200 A paths keep theirs. Load leakage behaves differently: give it a lower-impedance return and the original path's reading falls. The jumper test did not suggest a fault; it confirmed the induced-current mechanism.
Does one 4/0 equipment grounding conductor serve a whole parallel feeder?
No. NFPA 70, the National Electrical Code, 2023 edition, 250.122(F) requires each parallel run to contain its own equipment grounding conductor sized from Table 250.122 for the feeder's overcurrent device, and copper 4/0 is the Table 250.122 size for a device above 1200 A through 1600 A. One conductor remote from the phase groups raises the inductive reactance of the ground-fault path and weakens clearing. Confirm the final sizes with the authority having jurisdiction and the equipment documentation.
Can we stop the current by opening one end of the grounding conductor?
Never. The equipment grounding conductor exists to carry fault current until the overcurrent device clears, and the code prohibits breaking the grounding path to eliminate objectionable current. The current would largely return anyway through the tray, the ground buses and the building steel, and you would have removed the fault-clearing path while leaving the induced voltage in place. Remove the coupling instead.
Will the 15 A on each motor ground drop after we regroup the feeders?
Most of it should, because the motor circuits are 3-conductor-with-ground tray cable that is internally well grouped, so their near-identical 15 A readings point to the external field from the misgrouped feeders. Re-measure after regrouping at the same load. Any current that then tracks a single motor gets its own insulation-resistance test and a check for neutral-to-ground bonds or filter current.