Differential protection is the most selective scheme in a substation: it compares the current entering a protected zone with the current leaving it, and trips only when the two disagree. The basic Merz-Price circulating-current scheme does this with a single operating coil — and it works beautifully until a heavy external through-fault arrives. Then the very currents it is supposed to ignore can fool it into tripping a healthy transformer. The percentage biased differential relay was invented to fix exactly that. This post explains the three drawbacks of the simple scheme and how biasing cures them.
Drawbacks of the simple circulating-current (Merz-Price) scheme
A differential scheme responds to the vector difference Idiff = |i1 − i2| between the currents entering and leaving a protected zone. A simple (unbiased) relay trips whenever that difference exceeds a fixed pickup — but during a heavy external through-fault, three effects create a false spill current even though the protected apparatus is healthy:
| Drawback | What happens |
|---|---|
| 1. Unequal CT saturation & ratio errors | Even slightly non-identical CTs saturate differently at 15–20 times fault current, causing a large false spill current i1 − i2 to flow through the operating coil. |
| 2. Pilot-wire lead burden mismatch | Unequal pilot-wire lengths place unequal burdens on CT1 and CT2, so their secondaries no longer match. |
| 3. On-load tap changer (OLTC) ratio variation | Changing transformer taps alters the primary-to-secondary current ratio by ±10–15%, causing mal-operation even at normal load. |
The biased remedy: restraining coil and slope characteristic
Figure 6(a): Percentage biased differential relay schematic (87) and slope operating characteristic. (Diagram redrawn for this post.)
The circulating current is routed through a centre-tapped restraining (bias) coil (Nr) carrying the average through-current Ir = (i1 + i2)/2, while the operating coil (Nop) is connected to the centre tap and carries Idiff = |i1 − i2|. The relay trips only when the operating ampere-turns exceed a fixed percentage (K = Nr/Nop = 10%–40%) of the restraining ampere-turns:
|i1 − i2| ≥ K · [(i1 + i2)/2] + I0
As the external through-fault current rises, the restraining force grows proportionally and prevents false tripping. A genuine internal fault, by contrast, produces a large difference with little through-current, so it crosses the slope easily.
Worked example: how much spill can the relay tolerate?
Take a through-current of Ir = 10 A with a bias slope of K = 30% and a minimum pickup I0 of 0.5 A. The relay will only trip when:
Idiff ≥ 0.30 × 10 + 0.5 = 3.5 A
Now consider the OLTC problem: a tap change of ±10–15% alters the current ratio, so at a 10 A load the mismatch alone can produce a spill current of about 1.0–1.5 A. A simple unbiased relay with a low pickup would see that spill and trip a healthy transformer. The biased relay ignores it comfortably — 1.5 A is well below the 3.5 A threshold — yet it still trips fast for a real internal fault, where the difference current is many amperes. And when a 15–20 times through-fault drives both CTs toward saturation, the restraining current grows with it, so the threshold rises in step with the error.
Key points to remember
- The simple Merz-Price scheme mal-operates on heavy external faults because of unequal CT saturation (15–20× fault current), pilot-wire burden mismatch, and OLTC ratio variation (±10–15%).
- The biased relay adds a restraining coil carrying Ir = (i1 + i2)/2; it trips only when |i1 − i2| ≥ K · Ir + I0.
- Bias slope K is typically 10–40%: the bigger the through-current, the bigger the difference needed to trip.
- Internal faults produce a large difference with little through-current, so they still trip fast and selectively.
Based on Power System Protection & Switchgear.
