Sean McGuinness
@seanmacguinness.bsky.social
Electrical engineer/researcher at EPRI specializing in blackout investigations, blackout prevention, and power grid protection. Also grid simulations, relay configuration, coding, AI, and things. My views are my own and don't reflect my employer.
Not much more information available from Open Infrastructure Map but here's the link openinframap.org#14.34/-25.41...
October 15, 2025 at 8:29 AM
Not much more information available from Open Infrastructure Map but here's the link openinframap.org#14.34/-25.41...
This is a Google Maps view of Bateias 500kV substation. The line reactors are visible (rectangular boxes with fans on top). From open street maps it looks like the Ibiuna line where the fault happened is the top right
October 15, 2025 at 8:26 AM
This is a Google Maps view of Bateias 500kV substation. The line reactors are visible (rectangular boxes with fans on top). From open street maps it looks like the Ibiuna line where the fault happened is the top right
September 15, 2025 at 9:33 AM
11/11 In conclusion, we showed that there are cases where protection won't work, but there are relatively straightforward solutions available. We have performed a number of these studies for utilities over the years. Most of the recent studies focuser on blackstarting transmission grids from BESS.
September 13, 2025 at 11:42 AM
11/11 In conclusion, we showed that there are cases where protection won't work, but there are relatively straightforward solutions available. We have performed a number of these studies for utilities over the years. Most of the recent studies focuser on blackstarting transmission grids from BESS.
10/11 We didn't get to discuss it in the paper, but we also worked with HVDC Center to do HIL testing of a protection relays using their RTDS. This was time-consuming, but very helpful to verify the EMT simulation results and also verify the band-pass filters we added to the relay models worked.
September 13, 2025 at 11:41 AM
10/11 We didn't get to discuss it in the paper, but we also worked with HVDC Center to do HIL testing of a protection relays using their RTDS. This was time-consuming, but very helpful to verify the EMT simulation results and also verify the band-pass filters we added to the relay models worked.
9/11 Direct cosine filters are used in several commerical relays, but these let DC through which messes up the impedance phasor estimation. In reality, there's an additional processing block which filters out the dc, but it's not always documented in the manual. Your relay models can over-reach.
September 13, 2025 at 11:40 AM
9/11 Direct cosine filters are used in several commerical relays, but these let DC through which messes up the impedance phasor estimation. In reality, there's an additional processing block which filters out the dc, but it's not always documented in the manual. Your relay models can over-reach.
8/11 Relay modeling in EMT is challenging. Most of the signal processing is documented in the relay manual, but there's always a critical block or filter parameter missing. We did stochastic switching transformer onrush studies to verify inrush detection worked.
September 13, 2025 at 11:40 AM
8/11 Relay modeling in EMT is challenging. Most of the signal processing is documented in the relay manual, but there's always a critical block or filter parameter missing. We did stochastic switching transformer onrush studies to verify inrush detection worked.
7/11 We found we could energize surprisingly far into the grid while keeping fault levels high enough for distance and differential to work. We mostly used vendor-specific relay models, but I had to build a few models of the electromechanical relays.
September 13, 2025 at 11:39 AM
7/11 We found we could energize surprisingly far into the grid while keeping fault levels high enough for distance and differential to work. We mostly used vendor-specific relay models, but I had to build a few models of the electromechanical relays.
6/11 Line protection during soft-start can be complicated. If switch-on-to-fault is enabled using breaker status (52a), it won't activate. Even if it does, there probably won't be enough current for phase overcurrent to trip or voltage for distance protection to make a directional decision.
September 13, 2025 at 11:39 AM
6/11 Line protection during soft-start can be complicated. If switch-on-to-fault is enabled using breaker status (52a), it won't activate. Even if it does, there probably won't be enough current for phase overcurrent to trip or voltage for distance protection to make a directional decision.
5/11 Since the HVDC is controlling its ac terminal voltage, the fault current is roughly Ifault = Vterminal/Zgrid. As Zgrid includes the transformer impedance, the terminal voltage can need to be relatively high before the fault current exceeds line protection's min sensitivity (usually 100-400A).
September 13, 2025 at 11:39 AM
5/11 Since the HVDC is controlling its ac terminal voltage, the fault current is roughly Ifault = Vterminal/Zgrid. As Zgrid includes the transformer impedance, the terminal voltage can need to be relatively high before the fault current exceeds line protection's min sensitivity (usually 100-400A).
4/11 We also tried soft-starting the grid over about 2 seconds. Slowly ramping the ac voltage meant no inrush current and no harmonic distortion issues due to the low dV/dt.
September 13, 2025 at 11:39 AM
4/11 We also tried soft-starting the grid over about 2 seconds. Slowly ramping the ac voltage meant no inrush current and no harmonic distortion issues due to the low dV/dt.
3/11 With the hard-start approach, we found several cases of instability and harmonic resonance after energizing large transformers and long 400kV lines. The overvoltages and harmonic distortion caused the HVDC overvoltage protection to trip. The simple solution was picking up load before switching.
September 13, 2025 at 11:38 AM
3/11 With the hard-start approach, we found several cases of instability and harmonic resonance after energizing large transformers and long 400kV lines. The overvoltages and harmonic distortion caused the HVDC overvoltage protection to trip. The simple solution was picking up load before switching.
2/11 We looked at traditional blackstart methods where lines and transformers are energized one by one as well as soft-start where a lot of the grid was energised by raising the voltage from 0 to 1pu over about 2 seconds. All EMT simulations performed in Digsilent PowerFactory.
September 13, 2025 at 11:38 AM
2/11 We looked at traditional blackstart methods where lines and transformers are energized one by one as well as soft-start where a lot of the grid was energised by raising the voltage from 0 to 1pu over about 2 seconds. All EMT simulations performed in Digsilent PowerFactory.
5/9 We could energize surprisingly far into the grid while keeping fault levels high enough for distance and differential to work. We mostly used vendor-specific relay models, but I had to build a few models of the electromechanical relays.
September 13, 2025 at 11:27 AM
5/9 We could energize surprisingly far into the grid while keeping fault levels high enough for distance and differential to work. We mostly used vendor-specific relay models, but I had to build a few models of the electromechanical relays.
6/9 Relay modeling in EMT is challenging. Most of the signal processing is documented in the relay manual, but there's always a critical block or filter parameter missing.
September 13, 2025 at 11:26 AM
6/9 Relay modeling in EMT is challenging. Most of the signal processing is documented in the relay manual, but there's always a critical block or filter parameter missing.
7/ Some commerical relays use direct cosine filters, but these let DC through which messes up the impedance phasor estimation. In reality, there's an additional processing block which filters out the dc, but it's not always documented in the manual. This makes the relay model under or over-reach
September 13, 2025 at 11:26 AM
7/ Some commerical relays use direct cosine filters, but these let DC through which messes up the impedance phasor estimation. In reality, there's an additional processing block which filters out the dc, but it's not always documented in the manual. This makes the relay model under or over-reach
8/9 We didn't get to discuss it in the paper, but we also worked with HVDC Center to do HIL testing of a protection relays using their RTDS. This was time-consuming, but very helpful to verify the EMT simulation results.
September 13, 2025 at 11:24 AM
8/9 We didn't get to discuss it in the paper, but we also worked with HVDC Center to do HIL testing of a protection relays using their RTDS. This was time-consuming, but very helpful to verify the EMT simulation results.
9/9 In conclusion, we showed that there are cases where protection won't work, but there are relatively straightforward solutions. We have performed a number of these studies for utilities over the years. Most of the recent studies focuser on blackstarting transmission grids from BESS.
September 13, 2025 at 11:23 AM
9/9 In conclusion, we showed that there are cases where protection won't work, but there are relatively straightforward solutions. We have performed a number of these studies for utilities over the years. Most of the recent studies focuser on blackstarting transmission grids from BESS.
Most of these issues have well-established solutions (because extreme weather for your grid is often completely normal conditions somewhere else), but they are rarely cheap. Doing nothing can be more expensive in the long term. Good forecasts, diligent engineering, and brave decisions are needed.
September 10, 2025 at 5:26 PM
Most of these issues have well-established solutions (because extreme weather for your grid is often completely normal conditions somewhere else), but they are rarely cheap. Doing nothing can be more expensive in the long term. Good forecasts, diligent engineering, and brave decisions are needed.
Drought can also deplete hydro reserves leading to unusual power flows and stressed grids. This can result in load-shedding and more frequent frequency deviations (Brazil 2015) as well as greater regional power transfer risking angular instability.
September 10, 2025 at 5:20 PM
Drought can also deplete hydro reserves leading to unusual power flows and stressed grids. This can result in load-shedding and more frequent frequency deviations (Brazil 2015) as well as greater regional power transfer risking angular instability.