An 87T Trip on a GE Multilin 850 Took Down a Healthy GE Transformer—Here's What I Missed

In January 2024, a relay test contractor handed me a report I didn't exactly want. The conclusion was that our GE transformer—a 4 MVA, 13.8 kV to 480 V oil-filled unit from 1993—was healthy enough to keep serving the plant for years. The problem was the protection around it. The old electromechanical relays were obsolete, slow to operate, and they couldn't provide differential protection. The report recommended an upgrade to transformer differential 87T protection.

I approved the project in February, picked a GE Multilin 850 relay, and spent the next few months learning things the hard way. By September, a fault on a downstream feeder caused an 87T trip on the new relay and shut the entire plant down for eight hours. This is the story of how we got there, and the checklist I now keep in the relay cabinet.

Why the GE transformer needed a new relay

My title is plant electrical engineer, but I'm not a protection engineer. I can troubleshoot motor controls, program VFDs, and manage a five-person electrical crew. Relay coordination is something I do carefully, with help, because the margin for error is small.

Here's what made the transformer worth protecting better: it's a 4 MVA GE unit feeding the plant's 480 V switchgear. For that size transformer, a differential relay (87T) compares current entering the high side with current leaving the low side. If those currents don't cancel inside the relay's slope settings, the relay assumes an internal fault and trips the breaker. Overcurrent relays alone can't detect certain internal winding faults fast enough. The GE Multilin 850 gave us 87T differential plus backup overcurrent, event recording, and communication—all in one panel-mounted device.

The standard for current transformers, IEEE C57.13, defines accuracy classes like C100, C200, and C400. I knew that at the time. I just didn't take it seriously enough. That decision came back to bite me.

Mistake one: I ignored the difference between a contactor and a relay

During the April installation outage, we had to interface the GE Multilin 850's trip output with the existing 13.8 kV breaker trip circuit. The old scheme used a lockout relay, and the 850's output was supposed to energize that relay through an interposing relay. The drawing said "relay." The procurement guy ordered a definite-purpose contactor instead, because a contactor also has a coil and contacts and it was cheaper and in stock.

To be fair, contactors and relays do look similar. But the difference between a contactor and a relay matters when the circuit is a protection circuit. A contactor is built to switch power loads—motors, heaters, lighting. A relay is built to switch control signals reliably at low energy, with faster operation and more predictable pickup. In a 125 V DC trip circuit, the contactor we received had an AC coil. It chattered, then dropped out once the breaker's trip current started flowing. We lost two days waiting for the correct interposing relay.

That delay put us behind schedule, and the pressure made us sloppy later in the project. The contactor mistake cost about $3,200 in rework and labor, but its real cost was that it pushed us into a rushed commissioning.

A GE buck-boost transformer is not a general-purpose control transformer

In the same outage, we needed a 120 V station service feed for the new relay panel and test receptacles. Someone suggested using a spare GE buck-boost transformer from the warehouse. I assumed a transformer is a transformer and said fine.

That assumption was wrong. A GE buck-boost transformer is designed for modest voltage corrections, usually in an autotransformer connection—for example, raising 208 V to 240 V or dropping 240 V down to 208 V. It is not a substitute for an isolation transformer when you need to step 480 V down to a clean 120/240 V control supply. We discovered the issue during testing, and the buck-boost unit went back to the warehouse. The correct station service transformer arrived a week later. It was a $600 mistake in freight and labor, but it taught me to read the catalog page before ordering—not after.

The 87T trip on the GE Multilin 850

By late April, the GE Multilin 850 was installed and powered up. It passed secondary injection testing on the bench. We injected currents into the relay terminals, simulated internal faults, and watched it trip correctly every time. The relay was not the problem that day.

The problem was what we didn't verify outside the relay: the current transformers feeding it.

The transformer's high-side CTs on the 13.8 kV bushings were relaying-class CTs, I believe C400. The low-side CTs in the 480 V switchgear were older, lower-accuracy units—I want to say C100, but I'd need to pull the test report to be sure. I entered the ratios into the 850 correctly. What I didn't do was compare the CT accuracy classes and saturation curves.

In an 87T differential application, all CTs in the differential zone should have matched characteristics. If one CT saturates during a high through-fault while the other stays linear, the relay sees a false differential current. Modern digital relays can do a lot, but they can't invent current that a saturated CT didn't reproduce.

The moment of truth came on September 17, 2024, at 4:14 p.m. A cable fault on a downstream feeder—outside the transformer's differential zone—created a massive through-fault current. The feeder breaker cleared it in a few cycles, as designed. But the low-side CT in the 480 V switchgear saturated first. The GE Multilin 850's event record showed the LV current waveform flattening. The relay calculated a large difference between the HV and LV currents and operated its 87T element. The transformer's primary breaker opened, and the whole plant went dark.

The transformer was never in danger. The relay didn't malfunction. It was given bad data by mismatched CTs, and it did exactly what a differential relay is supposed to do with inconsistent current measurements: it tripped.

A generator, a surge protector, and a UPS lesson

Replacing the low-side CTs required another shutdown. We scheduled it for a Saturday in October, and because we had to de-energize the 480 V switchgear, the usual station power was dead. We brought in a Champion 8500 watt inverter generator to run temporary lighting, the relay test set, and a laptop running GE's relay software.

I set the laptop up on a cart with a cheap power strip that had a surge protector built in. A colleague asked if we had a UPS. "We have a surge protector," I said. He shook his head and explained the difference between a surge protector and a UPS: a surge protector clamps voltage spikes, but it does nothing during a sag or dropout. A UPS has a battery and inverter, so it carries connected equipment through brief power disturbances.

Sure enough, when an electric heater cycled on, the Champion generator's inverter output sagged for a moment. The laptop shut down in the middle of a settings upload. The surge protector just sat there. It had done its job for voltage spikes, but nothing protected us from a brownout. A small UPS would have prevented the whole episode. Since then, I won't work on a relay with generator power unless the laptop and test equipment are plugged into a real UPS.

What I check now before touching a GE transformer

I'm not going to pretend the September outage was the transformer's fault. It wasn't. The GE transformer is still running today, and the GE Multilin 850 is a solid relay. The failure was mine—I rushed a protection retrofit and skipped the parts that didn't fit into the outage schedule.

Here's the checklist I keep in the relay cabinet now:

1. Verify every CT nameplate: ratio, accuracy class, burden, and polarity. If a CT lacks relaying-class documentation, don't use it in an 87T differential zone.
2. Do an end-to-end test through the CT circuits, not just secondary injection at the relay terminals. The relay can be perfect while the wiring feeding it is wrong.
3. Read the schematic for control components. If the drawing says interposing relay, don't substitute a contactor. They are not interchangeable in protection circuits.
4. Match the transformer to its application. A GE buck-boost transformer is great for voltage correction, but not for an isolation control-power feed.
5. Bring a UPS when running sensitive relay software from a portable generator. A surge protector is not the same thing, no matter how many outlets it has.

The fundamentals of transformer differential protection haven't changed in decades: matched CTs, correct polarity, and a stable trip circuit. What has changed is that modern digital relays like the GE Multilin 850 give you event records, waveform capture, and communication—so when something goes wrong, you can see exactly what happened. That's a blessing, but only if you use the information to admit what caused it.

In our case, the cause was me. The relay did its job. The CTs didn't match. And the plant paid for it. I hope this checklist helps someone else avoid the same expensive lesson.