Finding a Chinese Equivalent for a Schneider or GE Numerical Protection Relay

Why relays are the software problem in this column

Every previous guide on this column deals with hardware where a parameter table, read carefully, is most of the truth. A numerical protection relay reverses that: the hardware is the least of it. The relay is its curves, element logic, event records and communications — all of which live in firmware and setting files. Replacing a Schneider MiCOM or a GE UR relay with a Chinese numerical relay is not a component change; it is migrating behaviour between two different design philosophies, and the system then behaves as the substitute’s engineers assumed, not as your original coordination study assumed.

The six verification fronts

1. Element mapping, element by element. Build a table: every protection element in your current settings — overcurrent stages, earth-fault logic, breaker-failure timing, voltage protection, synchronism — mapped to the substitute’s element set. Where a named element is missing or behaves differently (derivation, measurement window, restart logic), that is a finding, not a footnote.
2. Curve families and their grounded points. ‘Inverse curve per IEC’ is a category, not a curve. Obtain the full curve tables and compare them across the operating range, especially between pickup and 10× pickup — ground faults and arc-flash decisions live exactly there.
3. Analogue hardware. Input ratings (In, Un, frequency), burden, and the CT/VT secondary match. A substitution that changes the measurement chain by a small margin changes the accuracy class story of every downstream calculation.
4. Binary I/O and protocols. Contact counts, trip/interlock wiring, and the SCADA interface: IEC 61850 GOOSE/MMS profile, DNP3 mapping, or whatever your station actually runs. Name the profile revision and test a live telemetry exchange before award, not after.
5. Firmware and toolchain availability. Relay firmware revision parity and the setting software’s licensing regime: who holds it, how renewals run, and whether your team can still open the records five years from now. This line disqualifies more substitutions than any electrical parameter.
6. Commissioning and periodic tests. Witness FAT-style energy injection on the actual scheme: pickup values, timings at two or three operating points per curve, and record reading via the toolchain. A relay that cannot be maintained locally is a liability, whatever its price.

Keep it or switch it

• If your application is simple feeder overcurrent with no stored event-report expectations, an equivalent is defensible once element mapping and curve tables are verified.
• If the relay sits inside a networked protection scheme (differential, breaker-failure chains, GOOSE messaging between bays), the switch becomes an engineering project: re-perform the coordination study on the substitute’s behaviour or stay where you are.
• If the destination utility certifies relay models by name, that list binds before any parameter argument.
• If the vendor cannot commit in writing to the setting tool’s availability and data-export path, treat the whole offer as hold at best.

SpecSourcing’s free 48-hour initial screen returns a go / hold / walk signal on the substitute in your application; the paid Spec Match tier maps elements, curves and protocols against the installed base in writing.