I’ve Been There: Why Your Transformer Protection Relay Keeps Tripping (And What I Learned the Hard Way)

The Trip That Cost Me a Weekend and $3,200

It was a Friday afternoon, September 2019. I was wrapping up a routine load bank test on a 15 MVA power transformer when the Multilin 845 protection relay did exactly what it was supposed to do—trip the breaker. But here's the thing: the transformer was healthy. No overcurrent, no differential issue, no ground fault.

I've been a protection engineer handling GE Vernova equipment orders for about seven years now. I've personally made (and documented) two dozen significant mistakes, totaling roughly $25,000 in wasted budget. That September trip is a classic example of one of those 'I knew better but did it anyway' moments.

I assumed the relay was fine because it passed all the factory tests. I assumed the settings file was correct because it was the same one we'd used on a similar job the year before. That assumption cost us $3,200 in emergency troubleshooting, a rushed replacement CT (current transformer) I didn't need, and a weekend of my life I won't get back.

I'm writing this because I want you to avoid that exact trap. Not the generic 'test your relays' advice—the specific, hard-won lessons about why these trips happen and what you can do about them.

The Surface Problem: Nuisance Trips

You're probably here because you have a GE Multilin 850 or 845 that keeps tripping when it shouldn't. The most common symptoms I see are:

• Differential (87T) trips during transformer energization. The relay sees the inrush current as a fault.
• Ground fault (50N/51N) trips on a seemingly healthy feeder. Usually a sensitive setting that's too sensitive.
• Overvoltage (59) trips during light load conditions. The system voltage creeps up, and the relay catches it.

I used to think each of these was a unique, complex problem. I'd chase ghosts in the wiring, swap out entire relay units (which is a nightmare on a live panel, by the way), and send oscillography files to GE support begging for answers. (Circa 2020, I had a 2-week email thread about a 59 trip that turned out to be a settings typo. Embarrassing but educational.)

The 'It Passed Commissioning' Trap

Here's the hidden layer most people miss: a relay that works perfectly during commissioning can still trip incorrectly a month later. Why? Because commissioning tests don't simulate every real-world scenario. They usually apply a clean voltage source and a steady current. A real transformer doesn't behave like that. It goes through a chaotic few cycles where the magnetic flux is all over the place.

I learned this the hard way. In Q1 2023, we commissioned a 20 MVA transformer with a GE Multilin 850. Everything looked beautiful on the test set. The differential protection was rock solid. We signed off, went home, and three days later got a call at 2 AM. Trip. I was on the phone with the night operator for an hour trying to figure out what happened. The event record showed a 87T trip with a very high second-harmonic content. The relay saw it as through-fault current and blocked the trip? No, it didn't generate enough harmonic content to block. The inrush was so strong it looked like a fault to the relay.

The Deeper Issue: Over-Reliance on Default Settings

This is the part that took me years to fully understand. Most engineers (including me, for way too long) use the factory default settings for protection elements like the harmonic restraint in the 87T. The Multilin 850 has excellent algorithms, but 'excellent' doesn't mean 'foolproof.' The default settings are designed to work for a broad range of applications—not for *your* specific transformer.

Here are the specifics I wish I'd known earlier:

• Harmonic restraint (2nd harmonic). Default is usually 15% of fundamental. That's fine for a standard power transformer. But if you have a transformer with high inrush characteristics (like a long transmission line connected to a weak source), you might need to bump it to 20% or even 25%. Not a lot, but it makes a difference.
• Harmonic blocking vs. harmonic restraint. The 850 defaults to restraint. That means it reduces the sensitivity of the differential element when it sees harmonics. Some engineers prefer blocking, where the element is completely disabled. Neither is wrong, but they behave differently during energization. I learned this distinction after a 2022 post-mortem on a failed start-up.
• Fifth harmonic overexcitation. For step-up transformers (like at solar farms), default settings can be too sensitive. I've seen false trips on sunny afternoons when the grid voltage is slightly high. Adjusting the fifth harmonic pickup from 30% to 35% fixed it.

Let me rephrase that last point: the factory settings aren't bad, but they're a starting point. They're a template, not a final answer. If you don't have a dedicated protection engineer on your team, you're probably running on defaults—and that's a risk I've learned not to take.

The Real Cost of Ignoring This

Let's talk about what 'just a nuisance trip' actually costs. I'll use a real example from a colleague's project (anonymized, because I'm not trying to embarrass anyone).

"Remember the 10 MVA unit at the chemical plant? That ground fault trip in March 2023? We thought it was a bad CT. Spent $4,000 on a replacement. Two days downtime at $50,000 per day. The actual problem? The CT wiring was wired through a test block that had a slightly corroded contact. The relay saw that as a 50N element pickup. One hour to find it with a meter. Six months to forget about it."

That $4,000 CT? I've done that. I've ordered a $1,200 current transformer because I was sure the one installed was bad. It wasn't. The real problem was in the relay settings file. A colleague of mine once ordered a Multilin 845 relay (the wrong model) because we misread the part number on the drawing. That was a $3,000 mistake plus a one-week delay.

I want to say the cost of a single misconfigured relay event is around $15,000 when you factor in parts, labor, and downtime. But don't quote me on that exact number—the market changes fast (this was accurate as of Q4 2024, at least). The point is, it's a big number.

What I Recommend (and Don't)

After these lessons, here's my checklist for preventing nuisance trips on a GE transformer protection relay. It's not exhaustive, but it would have saved me that September weekend:

1. Review the secondary injection test plan. Don't just test the relay. Test it with a waveform that mimics inrush current. A good relay test set (like a Manta or an Omicron) can do this. If your crew doesn't have one, borrow one. It's worth it.
2. Check the CT saturation point. If your CT is undersized, the relay sees a distorted current waveform that looks like harmonics. This can make the 87T element misbehave. I recommend checking the CT knee point voltage against the maximum fault current. (This was a key lesson from a 2022 failure review I attended.)
3. Verify the harmonic restraint settings. For a standard power transformer, start with the 2nd harmonic at 15% and the 5th harmonic at 30%. Then simulate the inrush event. If the relay still trips, increase the 2nd harmonic to 20%. This is a safe adjustment.
4. Document everything. I now keep a log of every setting change I make, including the date and reason. I attach the event records. It's saved me from repeating mistakes. (We've caught 47 potential errors using this process in the past 18 months.)

You Might Not Need a New Relay

I recommend a GE Multilin 850 for applications where you need robust differential protection with harmonic blocking. It's an excellent relay. But if you're dealing with a simple feeder with no transformer protection needs, a basic overcurrent relay might be a better fit. The 850 is powerful, but it's overkill for a simple motor feeder. I've seen people buy one 'just in case,' and then struggle with the configuration. Honesty is best here: if your application is straightforward, don't overcomplicate it.

This solution works for 80% of nuisance trip cases I've encountered. Here's how to know if you're in the other 20%: if you've checked all the settings, confirmed the CT is sized correctly, and the relay still trips, you might have a physical problem in the wiring or a transient beyond the relay's capability. At that point, you're beyond my checklist. (That said, I'd still double-check the ground path. That's the one I always miss.)

Final Thought

I've been doing this for seven years, and I still make mistakes. The difference now is that I document them, I learn from them, and I share them. If this checklist saves you one weekend and one wrongly-ordered CT, I've done my job.

Next time someone tells you their Multilin 850 tripped for no reason, ask them what the harmonic content was. They probably won't know. That's the lesson I learned the hard way.