How system-based protection testing can unveil potential problems in complex protection schemes
Due to the increasing complexity of our electrical power systems, more complex protection schemes are used to protect transmission lines, transformers and other equipment. Simple, conventional protection testing using a single test set for injection of steady-state voltages and currents is not suitable to verify the correct behaviour of such advanced schemes.
Author(s): Thomas Hensler
The potential problems of these applications are more likely to be found in the complex interaction of individual relay settings than in a simple failure of a single protection element. A new approach with system-based protection testing in the field, where the relevant fault and operational scenarios are much more realistic and where the whole protection system is under test simultaneously, can unveil a lot of these potential problems during commissioning or re-commissioning of the relays.
In recent years our electrical power system networks have changed considerably. Due to distributed energy resources, power is fed into the network at many different locations and consumed far away from where it was originally generated. For example in Germany major wind power generation from the north of the country is also consumed far away in the south. Therefore, our power system networks are operated almost at their limits. This creates new challenges for protection relays.
A detailed test of the protection relays, which is compatible with the complex protection tasks, is mandatory, especially after changes within the network around the protection device. Conventional functional tests, which verify whether the protection settings are set correctly within the relay, are no longer sufficient in many cases. An application-oriented test, which executes realistic fault scenarios with the relay based on a dynamic simulation of the primary network, opens up new possibilities for efficient protection testing. This approach fulfils the new, complex requirements for protection testing and can unveil potential problems which were not previously identified. Examples for cases, which have been found so far are:
- No correct pickup of a relay element due to weak infeed conditions
- Pickup of a distance zone during pause time of an auto-reclose cycle with superimposed load flow
- Different pause time settings on auto-reclosing of relays on both line ends
- Overreach of distance protection on a short line into a power station due to extreme load angle at full load
- Non-selective trips with an open coupling bay for a fault in the dead zone of a bus bar protection
Different to conventional protection testing the injected test quantities are calculated directly from a model of the primary network topology. Within a new simulation software the topology of the protected primary power system is modelled and different fault cases can be executed with the test devices directly. The software is capable to control multiple time-synchronized test devices from one PC, so that simultaneous injection into multiple relays, which implement complex schemes using relay communication, is possible easily, even if the relays are located in different substations.
Additionally the software can adapt the simulated quantities in an iterative way to mimic a real-time closed-loop simulation even for testing of distributed protection systems in the field. In that way even complex relay logic can be observed and assessed easily.
Different to conventional functional protection testing, where usually steady-state signals are used for injection, with system-based protection testing the transient voltage and current signals are calculated directly from a model of the primary power system. Therefore, the primary network has to be modelled and the parameters of the network elements have to be entered into the test system software. On the other hand, the detailed settings of the protection devices under test are not relevant for the test (black-box approach).
First the topology of the primary network has to be defined, which is done within the test software using an easy to use graphical editor. The individual network elements are placed and connected with each other. Network elements are connected to nodes, whereas a node is either a bus bar or a simple node.
For the protection of overhead lines and cables, distance protection relays are used, where the fault impedance is derived from the short-circuit voltage and current measured at the relay location. The fault impedance measured is compared with the zone reaches, which are time graded to cover adjacent lines. More details on the principle of distance protection can be read in .
For protection concepts using distance protection for power lines, a correct setting of the zone reaches for a selective protection behaviour is essential. A simple functional test, where steady-state voltages and currents corresponding to the fault impedance are applied, does not really match the real situation. Mostly these steady-state voltages and currents are calculated from a simplified network model, which assumes a constant test current, and only a trip from no-load operation is checked.
For system-based testing, the parameters of the protected lines are relevant. Therefore, they can be entered into the software in different ways, as shown in Fig 1. The software then converts the values from one to the other variant.
Fig 1: Input of line parameters
The correct behaviour of distance protection relays is even more challenging for more complete topologies such as branched lines, parallel lines and intermediate infeeds. For testing, therefore, the topology can be modelled within the new test software and faults can be placed anywhere in the network.
Infeed conditions and short-circuit power
For a simulation of short-circuits in the primary system, the infeed conditions, or the short-circuit power of the infeeds, determine the magnitude of the short-circuit currents. The infeeds are modelled as voltage sources with a certain source impedance. This can be entered in different variants as shown in Fig 2.
Fig 2: Definition of infeed conditions
The short-circuit powers/short-circuit currents on network nodes (e.g. on a bus bar) can be verified within the software and compared with calculations or specifications from other sources (e.g. other offline network calculation software).
Double infeed, superimposed load flow and arc resistances
The protected line is mostly fed from both sides, a considerable load flow current is superimposed and different arc resistances at the fault location can occur. With a dynamic simulation based on a model of the protected line, the transient signals for voltages and currents for a realistic fault scenario are calculated as shown in Fig 3.
Fig 3: Simulation of a realistic fault on a line with double infeed, superimposed load flow and arc resistance
From the application software, this fault scenario can be executed directly with a test device on the protection relay and the trip behaviour can be checked and assessed. During practical tests within a substation, an overreach of a distance zone on a short line into a power station due to extreme load angle at full load has been detected and the protection settings had to be adapted accordingly.
Even more complex fault types are also possible, such as line-to-line-to-ground faults or double faults with different arc resistances. For multiple faults, an exact simulation of the time of the fault inception is necessary to see the adequate reaction of the protection relay in time.
Reaches for branched feeders and across parallel line segments
The time grading of distance relays for the upper zones is set to reach beyond the primary protected line and is used as a backup protection for the following line segments. For branched feeders, correct reaches for faults on the different parallel feeders are required, as shown in the topology in Fig 4.
Fig 4: Zone reaches for branched feeders
Load flow in the parallel feeders influences the reaches of the zones which can be simulated and verified during a test.
Another aspect is the zone borders for parallel line segments. The reaches are shifted considerably due to the different operating states of the individual line segments (line in operation or out of service). In this case (see Fig 5), the infeed situations on both ends are also relevant and have to be simulated correctly during the test.
Fig 5: Zone reaches for parallel line segments
An important point is the correct verification of the time grading for all relevant operational states, which is possible by using an application-oriented test immediately and in an efficient way. For parallel lines, the effect of the mutual coupling between the lines should also be considered. Using a fourth current output of the test device for the zero sequence current of the parallel line, a test of distance protection relays, which consider this in their algorithms, can be done. More details regarding mutual coupling is explained in .
Effect of intermediate infeeds
Reaches of upper zones are considerably affected for successor lines with intermediate infeeds. A correct test of the protection behaviour is only possible if this effect is modelled and simulated properly. Then the time grading across multiple line segments for a topology as shown in Fig 6 can be verified.
Fig 6: Multiple successor lines with intermediate infeeds
For the test, different infeed conditions, as they are expected in reality, can be simulated in a row and the trip behaviour of the protection relay can be assessed.
Time grading of distance protection across transformers
With distance protection relays used around power transformers, a time grading for the opposite side of the transformer is possible. Fault impedances and fault types on the other voltage level are changed considerably due to the transformer, at least for all vector groups with star-delta connections.
Therefore, a test of the protection device with a simulation of the transformer, as shown in Fig 7, is necessary so that a verification of the behaviour for faults on the high-voltage and low-voltage side of the transformer is possible.
Fig 7: Time grading of distance protection across a transformer
Testing of logic functions
Today protection relays use more and more logic functions to realise a selective protection concept for a specific protected power system. Individual protection elements are enabled or disabled according to the logic state from different sources of binary information.
Testing such logic functions is mandatory during commissioning, since the setup of the logic within the relay is usually one of the most difficult and error prone tasks. Again using an application-oriented testing approach, where the behaviour of the logic is modelled in the test and can be seen during test execution, is much easier and more comprehensible.
For a protection scheme using individual logic, multiple relays are often involved. Testing such a scheme as a whole, including all the communication links between the protection devices, provides much better test coverage and quality.
Therefore, a distributed test is necessary where all the protection devices within a scheme are under test simultaneously using multiple test devices capable of injecting voltages and currents in a time-synchronised way. More details on how the new test software can control multiple test devices in a time-synchronised way using GPS time sources and how to control remote test devices using simple network connections (and for longer distances using any Internet connection) can be read in [3,4].
Simulation considers reaction of protection devices – close loop
For a dynamic simulation of a fault or operation scenario, all the transients for currents and voltages are calculated upfront and injected using the test devices. For some scenarios, however, an adaptation of the simulation on the reaction of the protection system under test is required. For example, in reality a trip command from a protection relay causes a trip of a circuit breaker, which should cause a change of all currents and voltages within the simulation. A correct test of a breaker failure protection for a bus bar protection for example, or a complex scenario of an automatic reclosing is only then possible.
A dynamic simulation, which can react on circuit breaker commands (trip or close) of the protection system under test in real time, is called a real-time closed-loop simulation (hardware in the loop). Such a simulation is only possible with much effort for computing hardware and high costs in a laboratory environment. For a distributed test with multiple distributed test devices in the field, this is not possible because the delay times for the communication to the remote test devices are too long.
Iterative closed-loop tests for distributed tests in the field
An iterative approach for a closed-loop test is possible even for distributed test devices in the field. Therefore, the behaviour of a real-time closed-loop simulation is mimicked using multiple repetitions of individual test steps. A single scenario is simulated repeatedly where the reactions of all relays, which have been recorded during previous iterations, are considered correctly.
If we assume that all protection devices behave deterministically, that is they show the same reaction if we repeat the same scenario (some time tolerance is accepted by the application), using an automatic iteration of the test steps with the final test step we get exactly the same behaviour as with a real-time closed-loop simulation. The test is repeated again and again, until no further new reactions of the protection devices, not already recorded, are seen.
As an example, the following figures show some iterations of a test of an automatic reclosing from an end-to-end line protection system. Within the first iteration, the simulation cannot react on the instantaneous trips of both relays, so the fault persists during the whole simulation duration, as shown in Fig 8.
Fig 8: First iteration
Within the next repetition, the simulation considers the trip of the circuit breakers on both ends (after a certain circuit breaker delay time set within the software), so that the fault is cleared. Again the application cannot foresee that the relays will try to reclose after about 500 ms, which is shown in Fig 9.
Fig 9: Iterative test step after application of trip commands
This is repeated automatically until no further new reactions from the relays are seen. The final test step, including two (not successful) reclosing cycles, is shown in Fig 10 and is identical to a real-time closed-loop simulation.
Fig 10: Final test step after application of all trip and close commands
The iterative closed-loop simulation is controlled from application software which is capable of controlling multiple test devices, even remotely using a network connection. The software repeats all the test steps automatically, so that no manual interaction of the user is necessary. Using this approach, complex fault scenarios can be simulated, where the software adapts according to the reactions of the protection relays and without the need for the user to manually configure a complex sequence of events within the application.
For analysis and assessment of the results, the same tools are available as for a single-end test. All reactions of all relays are available as binary traces within the application and it is possible to measure and assess time delays between any event and any slope of a binary signal.
Using this approach, several issues with co-ordination of auto-reclosing for end-to-end line protection could be identified. A potential problem can occur if the pause times of the relays on both ends of a line are not set up accordingly (considering different circuit breaker delay times) so that a temporary line fault cannot be extinguished during the pause time.
An additional problem was found regarding the pickup of a distance zone during the pause time of an auto-reclose cycle with superimposed load flow, which could be simulated using the new dynamic testing approach in an end-to-end test.
Bus bar protection testing
Within the application software, the topology of the protected power system is modelled, the test quantities are calculated from the software in a consistent way and can be injected to the individual relay devices directly controlled by the protection engineer. Even complex tests, such as testing a distributed bus bar protection scheme consisting of multiple distributed field units, are easily possible, as shown in Fig 11.
Fig 11: Topology of a bus bar modelled within the software
A more detailed description of an application-oriented approach for testing a distributed bus bar protection can be read in  while in  a practical example of such a test in the field is given. Using this approach, non-selective trips with an open coupling bay for a fault in the dead zone of a bus bar protection could be identified during commissioning and could be fixed in the relay settings.
Application-oriented testing of protection relays offers new possibilities to verify complex cases with protection relays in our electrical networks. Using such tests means that not only are the protection settings in the relay checked, but the real behaviour of the protection can be seen. Therefore, even advanced protection functions can be tested and errors in the protection setup which were made during the design of the protection system can be detected. This is an efficient approach to guarantee correct and selective protection settings during commissioning in the field. By using an easy to use application software, improved quality for protection testing in a large number of situations is possible, and can be seen in Table 1.
Double infeeds, superimposed load flow and arc resistance
Switch onto fault
End-to-end test of teleprotection
Branched radial system and parallel lines
Automatic reclosing (even with single pole trip and close)
Line differential protection (even with multiple ends)
Breaker failure protection
Bus bar protection (even with multiple distributed field units)
Time grading over transformers
Isolated and compensated networks
All types of teleprotection
Mixed overhead lines and cables
Power swings and out-of-step blocking
Table 1: Use-cases for application-oriented testing
Additionally, the new test software can control multiple test devices in a time-synchronised way for a distributed test (and for longer distances using any Internet connection). Therefore, simple and efficient end-to-end tests for line differential protection or teleprotection schemes up to detailed tests of distributed bus bar protection systems are possible.
Using an iterative close-loop approach, tests which adapt to relay reactions such as trip and close commands in real time can be achieved, even for distributed cases, with simple portable test devices in the field.
- Ziegler G.: ‘Numerical distance protection, principles and applications’ (Publicis Corporate Publishing, 2011, 4th edn.).
- Pritchard C. Hensler T.: ‘Test and analysis of protection behavior on parallel lines with mutual coupling’. Australian Protection Symp., Sydney, 2014.
- Bastigkeit B. Pritchard C. Hensler T.: ‘New possibilities in field testing of distributed protection systems’. PAC World Conf., Zagreb, 2014.
- Bastigkeit B. Pritchard C. Hensler T.: ‘Testing distributed protection systems over the Internet cloud’. DPSP Conf., Copenhagen, 2014.
- Pritchard C. Hensler T.: ‘Test and verification of a bus bar protection using a simulation-based iterative closed-loop approach in the field’. Australian Protection Symp., Sydney, 2014.
- Fink F. Hensler T. Trillenberg F. et al.: ‘A system-oriented approach for testing a distributed busbar protection’ (in German “Systemorientierter Ansatz für die Prüfung eines verteilten Sammelschienenschutzes”), netzpraxis, Magazin für Energieversorgung, Heft 12, December 2014, pp. 40–46.