Managing Fault Current in Grid-Forming Systems

Have you ever tried to coordinate feeder relays with the substation transformer overcurrent elements and felt the math didn’t quite line up? It happens because the current seen on the transformer high side is not the same as what the feeder relays measure on the low side. The transformer’s turns ratio and winding configuration reshape the fault current before it reaches the high-side device. Here’s the step-by-step logic I personally use when checking coordination: 1) Understand the transformer connection A common North American distribution substation transformer is high side Delta / low side Yg. Don't forget: the Delta blocks zero sequence current from passing to the high side. 2) Know what each relay is measuring • Low-side feeder relays (phase/ground) measure positive, negative, and zero sequence current on the low-voltage base. • High-side phase overcurrent sees only positive and negative sequence current for a low-side line-to-ground fault because the delta traps I0. 3) Compare currents for the same fault For a single-line-to-ground fault on the feeder: • Feeder current: I(feeder) = I1 + I2 + I0 • High-side current: I(high side) = I1 + I2 • The feeder device responds to the full residual current, while the transformer protection is blind to I0. 4) Identify the tightest point of coordination Surprisingly, it’s not the LG fault. The toughest case is a LL fault near the substation: • Feeder side 50/51P sees about 87 % of the current it would see for a 3ϕ fault. • High-side transformer 50/51P sees nearly the full 3ϕ current because the delta winding passes positive and negative sequence unchanged. If you coordinate the feeder phase time-overcurrent 50/51P pickup and curve to clear before the high-side 50/51P for this LL case, you’ll generally maintain margin for all other fault types (including LG and 3ϕ faults). 5) Verify with actual curves Time-current curves on the low-side feeder relays and the high-side transformer protection must be compared using the converted current magnitudes each will experience. Only then can you be sure the feeder clears before the transformer trips for downstream faults. Real systems complicate this: zero-sequence compensation on feeder relays, different CT ratios, and relay curve shapes can all shift coordination. Questions for the community: • Have you seen feeders miscoordinate because someone forgot the delta blocks zero sequence? • Any lessons from real faults where the high-side transformer protection tripped first? I’d like to hear how others are refining these checks with today’s digital relays and modeling tools (ASPEN Inc., CYME, ETAP Software, EasyPower Software, SKM, etc). Comment or share your experience (or share this post if you found it valuable)!

Grid-forming inverters are one of the most discussed topics in power systems right now. But there’s a hardware constraint that doesn’t get enough attention outside technical circles. A synchronous generator — the machine that most power systems were built around — can deliver 3 to 7 times its rated current during a fault. That surge is what protection systems rely on to detect and clear faults quickly. A grid-forming inverter can deliver maybe 1.2 to 2 times its rated current. That’s not a software limitation. It’s physics — the semiconductor switches inside will burn out if you push more. So when a fault hits the grid, the inverter faces a conflict: protect itself, or keep behaving like the voltage source the grid needs. Two design strategies exist for handling this: Virtual impedance - the inverter adjusts its internal voltage reference to limit current while still acting as a voltage source. GFM behavior is preserved, but the response depends heavily on how well the impedance parameters are tuned. Mode switching - the inverter detects the fault and switches from voltage-source (GFM) to current-source (grid-following) mode. Current is controlled directly, but the GFM properties that make these inverters valuable in the first place disappear exactly when the grid needs them most. We’re adding more inverter-based generation every year. The grid codes are starting to require grid-forming behavior. But the fundamental tension between hardware limits and system needs hasn’t been resolved - it’s being managed. What approach are you seeing from manufacturers today — virtual impedance, mode switching, or something else?

Low fault current does not mean low protection impact. That may be one of the biggest mistakes engineers make in IBR heavy grids. Three reminders: • More MW does not guarantee more fault current • IBR fault response depends heavily on controls and current limits • Even modest fault current can still affect voltage, polarization, sequence components, and apparent impedance seen by protection Protection does not see MW. It sees voltage and current. Many protection schemes were designed assuming synchronous machine behavior. That assumption does not always hold for inverter based resources during faults. Safe takeaway In synchronous grids, machine physics mainly dominated fault response. In IBR plants, controls and current limits often dominate the response. #ProtectionEngineering #PowerSystems #IBR #Relays #GridProtection #ElectricalEngineering

How much fault current is needed from a Grid Forming (GFM) inverter for system reliability? How much MVA of GFM is needed for this behavior? Can studies be done in positive sequence environment to get an initial assessment? These were the questions we were attempting to answer in a recent study. The objectives were from a future planning perspective wherein additional IBRs have not yet been built/deployed. Rather, performance specifications were needed to be determined for these future IBRs. The study determined that transient current limits and steady state current limits plays a role in system fault ride through performance. Note that both current limits are active during the fault and the time duration for being active can be a determining factor. These results can give us more insights into details that may need to be added/considered while defining performance requirements for future IBRs. Lenna Lederman Marguerite Holmberg