I analyzed 600 SEL relays across 15 000 incidents. I gave 3 grades to SEL. Grade 1: Did every relay trip or not trip as designed?

I analyzed 600 SEL relays across 15 000 incidents. I gave 3 grades to SEL. Grade 1: Did every relay trip or not trip as designed??

I analyzed 600 SEL relays across 15 000 incidents in the American grid. I gave 3 grades to SEL:

Grade 1:?Did every relay trip or not trip as designed for each incident?

Grade 2:?Did all the word bits (elements) react as designed for each incident?

Grade 3:?Did the event report summaries data match the behavior of the waveforms on the event charts?

I analyzed at least 500 incidents events per each SEL model family. Not all the incidents necessarily led to a trip.

I did not do this exercise for fun. I had to convince several companies that their facilities' power systems modeling was significantly flawed. They could not run per design specs and had repeated significant electrical incidents primarily due to those flaws. They had no idea, and I needed to show them.?

I gathered all the data in SEL relays/devices and utilized their power simulation models. This article will not focus on how I proved that their facilities were flawed; I will write about SEL devices and my findings.?

I needed to show my expertise on their relays when I voiced my recommendation; therefore, I scrutinized almost every SEL device made with different firmware; the older relay and firmware were made in 2005.

300 relays were 351 models (feeder directional overcurrent):

- 351-5

- 351-6

- 351-7

- 351A

- 351S

The other 300 relays were a mix of other SEL relays/devices.

- 487V- Capacitor and protection control

- 387A- Current differential and overcurrent relay

- 311L- Line current differential

- 411L- Advanced line differential

- 587Z- High impedance differential

- 487B- Bus differential and breaker failure

- 421- Distance relay

- 451- directional overcurrent, protection, and bay control

- 387: Current differential and overcurrent relay

- 587- Current differential

- 751A- Feeder protection

- 751- Feeder Protection

- 2100- Logic processor

- 2411- Programmable automation controller

- 311C- Transmission protection

- No T401 nor T401L yet. I am currently working on it.

- Few others

Figure 1: Event report showing word bits to analyze

Grade 1:?Did the relays trip or not trip as designed?

For example, figure 1 above shows an event during a single-phase fault AG trip.

- Did the Trip word bit assert as designed?

- Did the Trip word bit assert within accuracy time shown on the manual specification?

I did this exercise for all the 15,000 incidents in which a trip occurred.

Figure 2 below shows a section of an SEL 351S user manual that shows all the accuracy. I scrutinized every SEL device against that accuracy specification and verified if they abide by their manuals.?

Figure 2: Relay manual showing some pick-up accuracies

Grade 2:?Did all the word bits (elements) react as designed?

For the same example in figure 1 shown above, word bits associated with trip equations were analyzed.

- Did all the word bits (apart from TRIP) assert as designed?

- Did all the word bits (apart from TRIP) assert within the accuracy time shown on the manual specification in figure 2?

- Should 51G assert? If yes, did it assert on time?

- Should 51GT assert? If yes, did it assert on time?

- Should 51P assert? If yes, did it assert on time?

- Should IN103 assert? If yes, did it assert on time?

- Should IN104 assert? If yes, did it assert on time?

- Should all those word bits stay asserted and never drop?

I did this exercise for all the 15,000 incidents. I did it both for a trip and a no-trip event.

Be aware that there are several hundred word bits in an SEL relay. For example, a fully loaded 351S relay has 777 word bits, with an average loaded 351S having 534 word bits.

I did not go through every word bit for every device in every incident. However, I picked 1 relay per family model, chose 1 of the worst-case incidents, and went through all the word bits.

Grade 3:?Did the event report summary data match the waveform data?

In figure 3 below, an event occurred, and the event report summary is shown on the right.

- Did the event AG match the fault type?

- Is the fault location accurate? If it shows $$$$, should it have?

I did this exercise for all the 15,000 incidents. I did it both for a trip and a no-trip event.

I did not consider evolving faults for this grade 3 since SEL wrote in the manuals that evolving faults might not accurately reflect on the event report summary.

Figure 3: Event report showing event summary

Final Results:

1- Grade 1 score: 100%

Every SEL relay that tripped during those 15,000 event reports tripped as designed and within specification accuracy times as shown in the manuals.

Every SEL relay did not send a false trip signal during those 15,0000 event reports.

2- Grade 2 score: 99%

99% of the word bits associated with the trip equations and event record equations asserted as designed and within specification accuracy times as shown in the relay manuals.

I found that 1% of the word bits should have asserted/deasserted quicker but did not.

3- Grade 3 score: 95%

95% of the event report summaries data matched the actual events on the field or matched the waveforms data from the event reports. 5% of the time, the event report summaries did not correspond with the actual field data.

Summary

---------------------------------------------------------------------------------------------------------------

Do not forget to subscribe and to share with your colleague. Remember, Knowledge is Power, and Power is Our Expertise.