Tiza Group

Home / Guides / Studies and design

Central or Distributed Interlocking: How to Decide

Criteria for choosing between interlocking in a central controller or spread across the bay IEDs, with a seven-feeder and coupler example.

By Shadai García, CEO and protection engineer at Tiza Group.

single busbar sectionalized by the bus couplerIEDIEDIEDIEDIEDIEDIEDIEDstation network: each IED publishes and receives via GOOSEbus coupler in the center; 3 feeders on the left and 4 on the rightone device fails (mid-cycle)central controllerblocked bays: 8 of 8distributed: the IED of feeder 3 failsblocked bays: 2 of 8average of the 8 devices: (7 × 2 + 8) / 8 = 2.75links: central 16 (8 + 8); distributed 14 subscriptionscentral scheme also requires a second controllerChainPositionsInterlockPermissionControl© tizagrp.comsingle busbar sectionalized by the bus couplerIEDIEDIEDIEDIEDIEDIEDIEDstation network: each IED publishes and receives via GOOSEbus coupler in the center; 3 feeders on the left and 4 on the rightone device fails (mid-cycle)central controllerblocked bays: 8 of 8distributed: the IED of feeder 3 failsblocked bays: 2 of 8average of the 8 devices: (7 × 2 + 8) / 8 = 2.75links: central 16 (8 + 8); distributed 14 subscriptionscentral scheme also requires a second controllerChainPositionsInterlockPermissionControl© tizagrp.com
Central or distributed interlocking: bays blocked when a device fails, on a sectionalized busbar with seven feeders and a bus coupler. Location in the system, operating principle and IEC 61850. Illustrative: actual values come from the nameplate and the manufacturer's manual.
Short answer

Interlocking in a central controller concentrates the logic in one place and makes changes easier, but a single failed device blocks the whole station. Distributed interlocking spreads the logic across the bay IEDs through GOOSE: a failure affects few bays, at the price of more distributed logic to test. The decision is made by measuring how many bays one device failure blocks, the response time and how the logic is changed and tested.

Short answer

There are two ways to solve interlocking between bays. In the central one, each bay sends its positions to a controller that decides and returns the permissives. In the distributed one, each bay IED receives by GOOSE the positions it needs and decides on its own. The design decision is made with three measures: how many bays are blocked by the failure of a single device, the response time and how much it costs to change and test the logic. In the example, the failure of the central controller blocks 8 of 8 bays; in the distributed scheme, the average failure of an IED blocks 2.75. Whichever option is chosen, the interlocking of the bay itself goes in the bay IED. For the catalog of interlocks by configuration, see substation interlocking; here the design criterion is covered.

What is interlocked where

Bay and station interlocking distinguishes two scopes. Bay interlocks use only information from the bay itself: the disconnector does not open or close with the breaker closed, and the earthing switch does not close on voltage. Station interlocks use information from other bays: the bus coupler, the busbars, the transfer disconnectors. This split decides the architecture: bay interlocks must keep working with communication down, and station ones are what is centralized or distributed.

The three options

Hardwired. The auxiliary contacts of one bay are wired to the logic of another. They do not depend on the network and are easy to understand, but each change is new wiring. They are used where there are few dependencies or where a utility criterion requires them.

Central. A station controller receives the positions, applies the logic and returns permissives. The logic is in one place, a change is made in one program and the test is a single one. Its limit: everything goes through it, so its failure affects everything that depends on it. The usual solution is to duplicate it, with its own switchover that also has to be tested. For the role of that device, see centralized protection and control.

Distributed. Each IED publishes its positions by GOOSE and subscribes to the ones it needs. There is no device on which everything depends, and a failure affects only the bays that use its data. In return, the logic is spread across several devices and a change in the substation requires touching several. The technical detail of the mechanism is in GOOSE interlocking (CILO and CSWI).

Criteria for deciding

CriterionCentralDistributed
Failure of a deviceBlocks everything that depends on it, unless duplicatedBlocks the bays that use its data
Logic changeOne programSeveral IEDs
TestOne logic, with all the inputsEach IED, plus the combinations
NetworkFewer crossed messages, but dependence on one deviceMore subscriptions, with their configuration
CostController and, if duplicated, its twinLogic in IEDs that are already there

Neither wins on everything. A small substation with few dependencies is usually solved with the distributed scheme without adding devices; one with many dependencies and frequent changes may prefer a duplicated controller. What decides is the number of bays one failure blocks and how often changes occur.

What must happen when something fails

Three behaviors have to be defined in writing before building. First, the state of each permissive when information is missing: the safe choice is to block the commands that depend on other bays. Second, local mode: the bay must be operable with its own interlocks even if the station is not there. Third, the override: bypassing interlocks must be controlled, with a key or a code, and leave a record of who used it and when. Without these three definitions the architecture is not finished.

How it is tested and how it is changed

Both solutions are tested the same way: each prohibited operation is set up and it is verified that the command is rejected, and each permitted operation is set up and verified to execute. It is repeated with communication cut and with each device out of service, to check the states the design defined. The difference is in the change: when the substation grows, in the central scheme one program is modified and what is affected is tested; in the distributed one the IEDs of the neighboring bays are modified and the new subscriptions are tested. In both cases, the version of the logic is stored with its date and owner, and the test is repeated in full if the change touches a safety interlock.

Worked example

Example data (assumed): a single busbar sectionalized by a coupler, with 7 feeders (3 on one side and 4 on the other) and the coupler, that is 8 bays. Dependency assumption: the interlocking of each feeder needs the coupler’s positions, and the coupler’s needs the positions of all feeders. Example times: bay position to controller 10 ms, controller cycle 100 ms, permissive back 10 ms; with GOOSE, 10 ms transfer and 10 ms logic. Example limit for the permissive: 500 ms.

  1. Failure of the central controller: all 8 bays lose the station permissive: 8 of 8.
  2. Failure of a feeder IED (distributed scheme): it affects that feeder and the coupler, which uses its position: 2 bays. There are 7 feeders, so 7 cases with 2 bays.
  3. Failure of the coupler IED: it affects the coupler and the 7 feeders that use its position: 8 bays.
  4. Average over the 8 devices: (7 × 2 + 8) / 8 = 22 / 8 = 2.75 bays, against 8 for the unduplicated central controller.
  5. Links: the central scheme uses 8 up and 8 down: 16. The distributed one uses 7 + 7 = 14 subscriptions.
  6. Time: central, 10 + 100 + 10 = 120 ms; distributed, 10 + 10 = 20 ms. Both are below 500 ms: time does not decide in this example; the failure does.

The example’s conclusion depends on its assumption: the coupler depends on all the feeders. With another split of dependencies the result changes, which is why it is calculated with the project’s real dependency table.

Checklist

  • Bay interlocks solved in the bay IED.
  • Dependency table between bays built.
  • Bays blocked by the failure of each device counted.
  • State of each permissive when information is missing defined.
  • Local mode of each bay defined and tested.
  • Control of the override (key or code) and its record defined.
  • If there is a central controller: redundant twin and switchover test.
  • If distributed: list of GOOSE messages with sender and receivers.
  • Communication cut tests in the acceptance plan.

References

  • Design criteria: our own, from protection and control engineering practice; no standard is transcribed.
  • Calculations and data: our own, with assumed data for the example.

Need this study?

We make the decision between central and distributed interlocking, with the dependency table and the logic of each bay, as part of our engineering design and studies service. We do it for projects in any country. Write to us.

Frequently asked questions

Can the two schemes be mixed in one substation?

Yes, and it is the most common arrangement: the interlocking of the bay itself (the breaker against its disconnectors and its earthing switch) is solved in the bay IED, and only what depends on other bays is shared through the central controller or GOOSE messages. That way the bay keeps its basic safety even if the network or the controller fails.

What should happen when communication is lost?

The design must say what state each permissive takes when information is missing. The safe option is to block the commands that depend on other bays and keep those of the bay itself, with a controlled, logged override procedure. That decision is tested by cutting communication, not left to assumption.

What is wrong with the central controller?

Nothing is wrong with it: it concentrates the logic and simplifies changes. Its weakness is that it is a single point of failure for every bay that depends on it, which forces it to be duplicated and its switchover tested. A redundant controller costs more and adds its own switchover logic.

WhatsApp