HV Tie-Line Ground Protection Near IBRs

A conventional transmission line is a tug-of-war between two big synchronous sources. Fault current pours in from both ends, and it lags the voltage by roughly the line angle at both ends. Line relays were designed around that symmetry. They compare currents between ends (87L), measure impedance (21), and judge direction from sequence quantities (32Q, 32V).

A gen-tie to a solar or BESS plant breaks the symmetry. The utility end is still strong. The collector end is an inverter-based resource. Its fault current is typically about 1.0–1.5 pu of rating, set by the OEM current limit, after a sub-cycle transient that can be higher (take the actual value from the OEM short-circuit model), and its phase angle comes from control firmware. In positive and negative sequence, the collector terminal looks like a programmable current source.

This module shows that the zero-sequence network is different. The inverters sit behind the MPT's delta LV winding, so they are invisible in HV zero sequence. The collector-end zero-sequence source is just the main power transformer's grounded-wye neutral, which is passive, linear and predictable. Good IBR tie-line ground protection is built on that one fact.

Assumed configuration: a 230 kV gen-tie to one YNd MPT (grounded-wye HV, delta 34.5 kV LV). The pad-mount arrangement on the collector (see the configurations table in Module 1.2) does not change the HV zero-sequence picture, because the MPT delta blocks it. A YNyn MPT, an autotransformer with a tertiary, or extra grounded-wye transformers on the HV bus do change it; the lab's "second grounded source" option shows one such case.

1. Line current differential (87L): non-homogeneous angles

87L compares the current entering each end. For an internal fault both currents point into the line, so the ratio $k = I_{remote}/I_{local}$ sits near the positive real axis. (The lab below takes the strong terminal U as local, so $k = I_C/I_U$.) With an IBR at one end, the angle of its current, especially its negative-sequence current, is whatever the controller chooses. If that angle swings far enough, k lands in the alpha-plane blocking region and the element restrains on an internal fault. A weak infeed with a small |k| is not the problem, because the blocking region excludes |k| < 1/R. A rotated infeed is.

2. Ground distance (21G) and current-based phase selection (CFIDS)

Ground distance elements must know which loop (AG, BG, CG) to trust. Many relays select it by comparing the angle of I₂ to I₀ (CFIDS), supervised by a negative-sequence overcurrent threshold. At the IBR terminal I₀ is passive but I₂ is synthetic. It may be too small to pass the 50Q supervision, or rotated so the sector logic points at the wrong phase. Even when the loop is correct, fault resistance is magnified by the infeed ratio and tilted by the remote current's angle, so the mho circle under-reaches. For ground loops the magnification is modest (about 1–4× in the lab's model) because the MPT makes C strong in zero sequence; it is the phase loops, fed only by current-limited I₁ and I₂, that see the large weak-infeed ratios.

3. Negative-sequence directional (32Q) and the 50QF trap

Normally 50QF/50QR supervision is set just above the standing negative-sequence current from load unbalance and untransposed lines, typically a small fraction of load current. One documented IBR application ended up with a much higher 50QF ≈ 1,228 A. The course could not confirm that application's rationale. A plausible one is security of the weak-end 32Q against the plant's own I₂ injection during external faults, which an IEEE 2800 plant produces at up to its current limit. Treat the number as a "what goes wrong if" example, not a recommendation: with the plant's I₂ capped by its current limit and its control law, the collector terminal cannot be relied on to produce that much 3I₂. When it doesn't, 32Q, 67Q and CFIDS all go quiet together.

Looking from the collector terminal into its own substation, the zero-sequence circuit is only the MPT's leakage impedance to ground. For any fault out on the line, the relay therefore measures

The measured z₀ does not depend on the fault location, the fault resistance or the inverter firmware. A voltage-polarised zero-sequence directional element (32V) therefore has a fixed, predictable operating point. Because 3I₀ is not limited by the inverter's MVA rating, its supervision (50GF) can be set far below 50QF. The example methodology uses 600 A.

• VFIDS: select the faulted loop from phase-voltage depression rather than the I₂/I₀ angle, so firmware cannot mislead it.
• 67G: operate when 50GF picks up and either 32V or 32Q declares forward. 32Q still helps when the plant does produce a good I₂.
• Z0F threshold: it must sit above every forward value z₀ can take and below every reverse value, with margin. Two approaches are in use. Some relays' automatic setting rules put Z0F at a positive value, roughly half the line's |Z0L| (about +36 Ω here), relying on reverse faults measuring a large positive z₀ (about +87.5 Ω for a bus fault behind C). This course's earlier, more conservative choice was half of −|Z0,MPT| (−8.8 Ω), close to the forward operating point. Check your relay's instruction manual for its actual rule. A positive Z0F relies on the remote terminal supplying the reverse-fault I₀, so it is secure only while U is in service; the lab's reverse check assumes it is.

What eats the margin is anything that moves the forward z₀ towards zero or the reverse z₀ down: other zero-sequence sources behind C (a second MPT or a grounded tertiary; grounded-wye station-service transformers too, though a small one has hundreds of ohms of Z0 and moves z₀ only slightly), CT and VT angle error, and mutual coupling. A second MPT (or a similarly stiff source) in parallel halves the impedance behind C, so line faults measure about −8.8 Ω. The forward margin of a −8.8 Ω Z0F then falls from about 2:1 to about 1:1, which is no margin at all: in the lab it fails by a few hundredths of an ohm, and any real angle error would make it fail outright. Try it in the lab.

Transmission corridors are crowded. If the tie-line shares towers or a right-of-way with another circuit, a ground fault on that circuit induces a zero-sequence EMF along the tie-line. The induced EMF is inside the line, so both terminals see zero-sequence current flowing in from their own source. Both z₀ values go negative and both 32V elements declare forward. To a POTT scheme built only on 32V, that looks exactly like an internal fault. (The lab models only this induced component; in a connected grid a conductive component also flows, so the real magnitude needs a coupled-network study.)

The first two rows pull against each other in this example. Covering every fault up to 40 Ω (challenge task 3) needs 50GF at or below about 509 A, the smallest 3I₀ at C (a 40 Ω fault next to U). The worst-case induced 3I₀ at Z0M/Z0L = 0.6 is about 712 A, so no single 50GF pickup is both dependable and secure. Here 32Q supervision at U is mandatory, not optional.

The strong terminal has its own problems, and they come from the same MPT.

• Zone 2 reaches into the plant. On a short gen-tie, a 120 % Zone 2 at U reaches through the line and into the MPT (with this module's impedances, about a quarter of the way through it) and, on very short lines or with minimum-reach rules, possibly the 34.5 kV collector. The plant's small infeed hardly reduces that reach. Coordinate it with the MPT 87T and backup protection and with the collector bus and feeder protection, or time-delay it accordingly.
• The MPT is a ground source on the utility system. For line ground faults the MPT neutral supplies part of 3I₀, so U sees less than the total. For faults elsewhere on the utility system, the YNd MPT feeds I₀ back through the line. The utility's ground-relay coordination must include the MPT as a zero-sequence source, and must be checked with the MPT in and out of service.
• Transformer-terminated lines. For very short gen-ties, some designs use one differential zone covering the line and the MPT (for example 87L with the transformer inside the zone), which can remove the MPT HV breaker. It is a design choice to agree with the transmission owner.

IEEE C37.246 (guide for protection of transmission-to-generation interconnections) covers these interconnection questions, and the IEEE PES technical report on IBR impacts on transmission protection (cited by this course as PES-TR81; verify the report number and edition) covers the IBR-specific behaviour.

With POTT both ends normally trip for a gen-tie fault. If the C breaker fails to open or is slow, the plant is left islanded on the line when U opens, and U must not reclose onto a line the plant may still be energizing. The usual safeguards are dead-line supervision (27 on a line-side VT) or synchronism check (25) at U, a direct transfer trip from U to the plant, and a reclose dead time longer than the plant's worst-case trip or cessation time. These must be coordinated with the plant's anti-islanding and ride-through settings (IEEE 2800 and PRC-029-1 keep it online for faults elsewhere). A failed reclose is also a second voltage dip for other IBR plants nearby (this plant is already disconnected), and it counts toward their consecutive-deviation rules in Module 3.2. Single-pole tripping and reclosing with IBR plants is an active area: how a converter behaves during the open-pole interval is OEM-specific, so check with the OEM before applying it.