Fixing Differential Relay Tripping on 15 MVA Yy0 Power

1. Problem Overview

A 15 MVA, 132/11 kV, vector group Yy0 power transformer fitted with three Reyrolle type C21 electromechanical duo-bias differential relays trips at approximately 15 % of rated load. The tripping reappeared after a fire-damaged LV CT panel was replaced. The original LV CTs were ratio 800/1 A; the replacement CTs are Alstom type VMXspec BS/1S, ratio 1200/0.577 A. Siemens 4AM22 20-7AA T40/A interposing CTs (ICTs) are wired in star/delta on both sides. Earlier schemes that combined an HV ICT at 6/12 ratio and an LV ICT at 7/8 ratio allowed operation up to about 50 % of rating before misoperation resumed. The HV side is solidly grounded, the LV side is resistance grounded, and a restricted earth-fault (REF) scheme is connected to the HV neutral CT.

Failure to balance the through-current phasors on both sides of the differential element is the dominant cause. The error grows with load because any residual mismatch is multiplied by load current. At 15 % load a 0.1 A mismatch appears as 0.1 A operating quantity; at full load the same mismatch appears as 0.66 A, well above the relay bias characteristic.

2. System Data and Base Quantities

For the LV CT, the nominal ratio is 1200/0.577 A. The 0.577 A figure equals 1/sqrt(3), which is the conventional way of rating a CT intended for delta connection so that the line current at the relay terminals equals the primary line current divided by the primary ampere rating (i.e. 787.3/1200 = 0.656 A).

3. CT and ICT Ratio Mathematics

3.1 Ideal balance condition

For a Yy0 transformer the phase shift between HV and LV is zero, so vector compensation is not required. The only task of the ICTs is to equalise the secondary currents of the main CTs. The condition is:

I_HV_CT(sec) x N_HV_ICT = I_LV_CT(sec) x N_LV_ICT

where N is the ICT turns ratio (primary/secondary). With star HV CTs and delta LV CTs both quantities at full load equal 0.656 A, so 1:1 ICTs on each side give perfect balance.

3.2 Worked example with the failing scheme

Original failed scheme: HV ICT 10:10 (1:1), LV ICT 26:15 (1.733:1) on windings 10:10 and 26:15 respectively.

A 1.733:1 ratio was applied where unity was required. The differential operating current therefore rose in proportion to load: at 15 % load the relay already saw roughly 0.17 A, comparable to the C21 operating threshold, which is why the relay tripped. The 26:15 tap was almost certainly chosen for a Dy1 or Dy11 vector group where 1/sqrt(3) ratio correction is required. For Yy0 it is the wrong connection.

3.3 The "partially working" scheme

The 6/12 HV ICT and 7/8 LV ICT were reported to carry about 50 % of rated load. The computed balance factor is:

• HV side: 0.656 x (12/6) = 1.312 A at the relay
• LV side: 0.656 x (8/7) = 0.750 A at the relay
• Imbalance: (1.312 - 0.750) = 0.562 A

This imbalance is itself large enough to operate the relay at full load, but the relay may have tolerated it at 50 % load because the operating current is then only 0.28 A. CT saturation at higher current and through-fault currents will aggravate the imbalance further. The "stability up to 50 %" is a coincidence of threshold, not a sound design.

4. Root Cause Analysis

5. Recommended Differential Scheme for Yy0

For a Yy0 transformer the cleanest and most defensible scheme is:

1. Main HV CTs: 100/1 A in star.
2. Main LV CTs: 1200/0.577 A in delta (or 1200/1 A in star with a 1:1.732 ICT).
3. HV ICT: D-D connected 1:1 (windings chosen for unity, e.g. 10:10 on the 4AM22 20-7AA).
4. LV ICT: omitted entirely when the LV CTs are already in delta, since the delta connection of the main CTs both provides the correct magnitude (line current at the relay equals primary line/1200) and removes the zero-sequence path needed for external HV earth faults.
5. Where REF is required on the HV neutral, the HV ICTs must remain D-D (not Y-D), because a Y-D ICT introduces a neutral that interferes with the REF neutral CT connection.

6. Step-by-Step Procedure to Re-balance the Scheme

1. Confirm the vector group from the transformer nameplate. Do not trust a previous scheme; some Yy0 units are in fact YNyn0 internally and require a star-point CT for REF.
2. Record main CT nameplate data on both sides: ratio, accuracy class, knee-point voltage Vk, secondary resistance, polarity mark.
3. Compute the required ICT ratio using Section 3.1. For a Yy0 transformer the answer is normally 1:1 on both sides if the main CTs are connected symmetrically (HV star, LV delta at 1200/0.577).
4. Select ICT taps on the Siemens 4AM22 20-7AA. The T40/A variants are typically tapped at 10, 15, 20, 26, 30, 40, 50 and 80 turns. For 1:1 use 10:10 (or 15:15, 20:20, 26:26 etc., as available on each winding). Mark the chosen taps on the schematic.
5. Connect the HV ICT in delta-delta so that any zero-sequence current from an external HV earth fault circulates in the delta and does not reach the relay. This is the safest topology for the Yy0 / solidly grounded HV case.
6. Connect the LV ICT (or omit it) per Section 5. If a 1.732:1 ratio is required, use 26:15 (1.733) on the 4AM22; this is intended for Dy compensation, not for Yy0, and must be reviewed before use.
7. Verify polarity by injecting a small AC current through one phase CT and observing all six relay inputs (HV a, b, c and LV a, b, c). All three HV phasors must be equal in magnitude and 120 degrees apart, and likewise for the LV phasors. The HV and LV phasors for the same primary phase must be in phase (Yy0 has zero displacement).
8. Apply a balanced three-phase injection at 50 % of rated secondary current on each side. Operate quantities on all three relays must be below 5 % of through quantity, and bias quantities must be equal on operate and restrain coils.
9. Step the load from 0 to 100 % in 25 % increments. Record operate and bias quantities at each step. A balanced scheme shows operate quantity proportional to the small CT error curve (typically below 0.05 A) and bias quantity proportional to through load.
10. Apply an external single-phase-to-earth fault on the HV side (test source or staged fault). Confirm that the HV delta ICT suppresses the zero-sequence current at the relay terminals, and that REF operates correctly for an internal HV winding earth fault.

7. Verification Tests

8. Common Pitfalls and Field Caveats

• Reusing the previous scheme's ICT taps after a CT swap. The taps were correct for 800/1 star CTs; they are not correct for 1200/0.577 delta CTs. Always recompute.
• Mixing 30 degree shifts in opposite directions. If one delta is leading (Dy11) and the other lagging (Dy1), the through currents add instead of cancelling. For a Yy0 transformer this is invisible on the nameplate but visible at the relay as a 3:1 ratio error.
• Polarity reversal in the ICT. A primary/secondary swap on a 4AM22 produces a 1:N error in addition to any phase error. Verify polarity with a known injection on every winding.
• ICT ratio chosen to satisfy the bias characteristic but not the through-current balance. The C21 duo-bias design has a through-current stability feature that masks small ratio errors at low load. Errors appear as load increases.
• Star-point CT for REF wired in the wrong star. If the HV ICTs are reconfigured as Y-D and a star point is created, the REF neutral CT must remain on the main CT star, not the ICT star.
• Floating-delta connection on unused ICT windings. Leaving unused 4AM22 windings open-circuited can cause high voltages during through faults. Tie the unused windings in a closed delta or short them before energising.
• CT secondary loop open during a switching event. A blown CT link or open test block produces a huge voltage and saturates the CT, with subsequent misoperation on re-energisation. Use shorting-type terminal blocks.
• CT polarity mark on the Alstom VMXspec. The replacement CT is a moulded-resin block; the polarity is often marked with an arrow or an embossed letter. Confirm with a polarity check before energising.
• Residual flux in the new LV CTs. A new CT may have residual magnetism from the factory test. Demagnetise by applying a slowly decaying AC voltage to the secondary before commissioning, otherwise ratio errors of 1 to 3 % can appear.

9. Sizing Reference - Worked Example from Nameplate

For a different transformer rating the balance procedure is the same. Use the three formulas:

1. Primary line current: I_line = S_kVA / (sqrt(3) x V_LL_kV x 1000)
2. CT secondary magnitude at full load: I_sec = I_line / CT_ratio
3. Relay current: I_relay = I_sec x N_ICT (for star CTs) or I_relay = I_sec x N_ICT x sqrt(3) (for delta CTs at the corner)

Set I_relay_HV = I_relay_LV and solve for the required N_ICT. Round to the nearest 4AM22 tap combination. For a Yy0 transformer the result is normally 1:1 on both sides when the main CT ratios are chosen to satisfy the rule that the HV secondary and the LV secondary at the relay terminals are equal at full load.

10. Reference Standards

Designs, tests and on-site commissioning should be cross-checked against the following documents:

• IEC 60255-1 - Measuring relays and protection equipment - Common requirements
• IEC 60255-118-1 - Measuring relays and protection equipment - Part 118-1: Power system protection with digital interfaces
• IEEE C37.91 - Guide for Protecting Power Transformers
• IEC 60076-1 - Power transformers - General
• IEC 61869-2 - Instrument transformers - Part 2: Additional requirements for current transformers

11. FAQ