Directional overcurrent
The key idea
On a radial feeder the current has only one path, so its size is enough to protect the line. Close a ring or add a second source, and that changes. The same relay then sees nearly the same current for a fault on either side of it. Only the angle tells the relay which side. The relay measures that angle against a voltage.
The idea
On a radial feeder this problem does not arise. Power leaves the source and travels outward. Any current above the setting is a fault somewhere ahead. An overcurrent relay that measures only amps has everything it needs.
Close that feeder into a ring, or connect a second source at the far end. The statement above stops being true. Current can now reach a fault from either direction. A relay in the middle of the line sees current for a fault on the section it protects. It also sees current for a fault behind it, on the neighboring section. The two currents are often within a few percent of each other in size.
The relay measures only a magnitude, and the magnitude does not name the faulted section. The angle does name it. Fault current is set almost entirely by reactance. A faulted line is inductive, and its resistance is small by comparison. So the current lags the voltage that drives it by close to 90°. Give the relay a voltage to measure that lag against, the polarizing voltage, and the angle of the current becomes a usable fact.
Now move the fault to the other side of the relay. The current in that conductor physically turns around. It does not shrink or grow. It reverses. On a phasor diagram the current turns through 180°.
A directional element draws a straight line through the origin of that phasor plane. It then tests which side of the line the current falls on. The orientation of the line comes from the maximum torque angle (MTA). The MTA is the current angle that the element is most sensitive to. Engineers set the MTA among the angles that real faults produce. The element allows a trip for current on the forward side of the line, and it blocks everything else.
Direction does not replace the overcurrent element. It only gates that element. The relay still needs enough current, and it still waits for its graded time. The directional check only decides whether this relay should act on the fault.
Try it
Slide the fault across the relay. The current barely changes in size, but the verdict changes completely.
through the relay: 2.50 pu at −90° vs V
FORWARD — relay may trip
The model uses two 1∠0 pu sources and pure reactance: 0.15 pu of source, 0.2 pu of line, and 0.15 pu of source. The relay sits 30% of the way along. A fault at 0.80 gives the relay 2.50 pu. A fault at 0.10 gives it 2.22 pu. The two currents are 11% apart in size, and they have opposite meanings. A ring needs direction for exactly that reason.
Why it matters
- A closed loop makes direction necessary. A radial feeder never needs direction. Once a ring closes, or a generator connects at the far end, every relay inside the loop must know which direction it protects.
- Grading assumes an order that exists in one direction only. A graded set of curves runs from the fault back to the source. When current can arrive from either end, you grade two sets, one for each direction. The directional element then picks the set that applies.
- A blocked relay still works correctly. The relay nearest a fault on the neighboring section still measures thousands of amps. Without direction it trips and disconnects a healthy section with the faulted one. The outage then spreads past the zone that should contain it.
- The reference is the weak point. A directional element depends on a voltage measurement. A blown VT fuse removes that reference, and the element can no longer decide the direction. Relays therefore carry fuse-failure supervision. Engineers also choose a polarizing voltage that the fault does not collapse.
The math, if you want itOptional — the page reads completely without it
The element compares the current's angle against its characteristic and operates on the forward half of the plane:
directional operate criterion
cos( θI − MTA ) > 0
The relay measures θI from the polarizing voltage. The MTA is the angle of highest sensitivity. The boundary sits 90° to each side of the MTA, so the operate region is exactly half the plane. The widget uses an MTA of −45°.
The current lands near −90°, because the reactance of a faulted line dominates its resistance:
fault current angle
θI = −arctan( X / R ) → −90° when X ≫ R
The reversal is a full 180°, and not an approximation. For a fault beyond the relay, the relay carries the contribution of the left source, and that current flows to the right. For a fault behind the relay, it carries the contribution of the right source, and that current flows to the left. The conductor and the measuring point stay the same, and only the sign changes:
the two cases at the relay
θI = −90° forward · θI = +90° reverse
Two simplifications are worth naming. The widget models the network as pure reactance, and it polarizes against a single voltage reference. Every current therefore comes out at exactly ±90°.
Real relays polarize a phase element from a quadrature voltage, the line-to-line voltage of the other two phases. A close-in fault collapses the voltage of the faulted phase itself, and that would leave the element with no usable reference. Real networks also have resistance, so the fault current angle stops short of 90°. Engineers therefore set the MTA near the real fault angle rather than at −90° exactly.
See it in Phasor
Phasor knows the topology, so it knows which relays sit inside a loop. The overcurrent elements of those relays need directional supervision. Set the direction and the MTA on each element. The fault study then reports the current that every relay sees, with its angle, for a fault on either side of it.