Home / Technical Articles / Protection / The Essentials Of Numerical Relays, Their Features And Important Considerations

Introduction to numerical relays

The distinction between digital and numerical relays is particular to Protection. Numerical relays are natural developments of digital relays due to advances in technology. They use one or more digital signal processors (DSP) optimised for real time signal processing, running the mathematical algorithms for the protection functions.

The Essentials Of Numerical Relays, Their Features And Important Considerations
The Essentials Of Numerical Relays, Their Features And Important Considerations

The continuing reduction in the cost and size of microprocessors, memory and I/O circuitry leads to a single item of hardware for a range of functions. For faster real time processing and more detailed analysis of waveforms, several DSPs can be run in parallel.

Many functions previously implemented in separate items of hardware can then be included in a single item.

Ok, let’s start with introduction to numerical relays, then explaining their additional features and at the end to discuss important considerations regarding software, data management, etc..

Table 1 provides a list of typical functions available, while Table 2 summarises the advantages of a modern numerical relay over static relay equivalents.

TABLE 1 – Numerical distance relay features

Distance Protection – several schemes including user definableOvercurrent Protection (directional/non-directional)
Several Setting Groups for protection valuesSwitch-on-to-Fault Protection
Power Swing BlockingVoltage Transformer Supervision
Negative Sequence Current ProtectionUndervoltage Protection
Overvoltage ProtectionCB Fail Protection
Fault LocationCT/VT Supervision
AutorecloseCheck Synchronisation
CB State MonitoringCB Condition Monitoring
Broken Conductor DetectionUser-Definable Logic
Fault/Event/Disturbance recorderMeasurement of Power System Quantities (Current, Voltage, etc.)

TABLE 2 – Advantages of numerical relays over static relays

Several setting groupsWider range of parameter adjustment
Communications built in (serial, Ethernet, teleprotection, etc.)Internal Fault diagnosis
Power system measurements availableDistance to fault locator
Disturbance recorderAuxiliary protection functions (broken conductor, negative sequence, etc.)
CB monitoring (state, condition)User-definable logic
Backup protection functions in-builtConsistency of operation times – reduced grading margin

Figure 1, Figure 2 and Figure 3 show typical numerical relays, while Figure 4 and Figure 5 show typical numerical relay boards.

First generation transmission numerical relay (1986)
Figure 1 – First generation transmission numerical relay (1986)
First generation distribution numerical relay
Figure 2 – First generation distribution numerical relay
Typical modern numerical relay
Figure 3 – Typical modern numerical relay

A numerical relay has the functionality that previously required several discrete relays, therefore the relay functions such as overcurrent or earth fault are referred to as ‘relay elements’.

Each relay element is in software so with modular hardware the main signal processor can run a vast variety of relay elements.

Numerical relay processor board
Figure 4 – Numerical relay processor board
Numerical relay redundant Ethernet board
Figure 5 – Numerical relay redundant Ethernet board

The argument against putting many features into one piece of hardware centres on the issues of reliability and availability. A failure of a numerical relay may cause many more functions to be lost, compared to applications where different functions are implemented by separate hardware items.

Comparison of reliability and availability between the two methods is complex as inter-dependency of elements of an application provided by separate relay elements needs to be taken into account.

With the experience gained with static and digital relays, most hardware failure mechanisms are now well understood and suitable precautions taken at the design stage. Software problems are minimized by rigorous use of software design techniques and the ability to download updated software.

Practical experience indicates that numerical relays are as reliable as relays of earlier technologies. Modern numerical relays will have comprehensive self monitoring to alert the user to any problems.

Go back to main index ↑

1. Hardware Architecture

The typical architecture of a numerical relay is shown in Figure 6. It consists of one or more digital signal processors, some memory, digital and analogue input/output (I/O), and a power supply.

Where multiple processors are used, one of them is a general controller of the I/O, Human Machine Interface (HMI) and any associated logic while the others are dedicated to the protection relay algorithms.

Typical numerical relay hardware architecture
Figure 6 – Typical numerical relay hardware architecture

By organising the I/O on a set of plug-in printed circuit boards (PCBs), additional I/O up to the limits of the hardware  / software can be easily added. The internal communications bus links the hardware and therefore is a critical component in the design.

It must work at high speed, use low voltages, yet be immune to conducted and radiated interference from the electrically noisy substation environment.

Excellent shielding of the relevant areas is therefore required. Digital inputs are optically isolated to prevent transients being transmitted to the internal circuitry. Analogue inputs are isolated using precision transformers to maintain measurement accuracy while removing harmful transients.

Additionally, the input signals must be amplitude limited to avoid them exceeding the measurement range, otherwise the waveform is clipped, introducing harmonics.

See Figure 7.

Clipping due to excessive amplitude
Figure 7 – Clipping due to excessive amplitude

Analogue signals are converted to digital form using an A/D converter. The cheapest method is to use a single A/D converter, preceded by a multiplexer to connect each of the input signals in turn to the converter. The signals may be initially input to several simultaneous sample-and–hold circuits before multiplexing, or the time relationship between successive samples must be known if the phase relationship between signals is important.

The alternative is to provide each input with a dedicated A/D converter and logic to ensure that all converters perform the measurement simultaneously.

The frequency of sampling must be carefully considered, as the Nyquist criterion applies:

fs ≥ 2 × fh


  • fs = sampling rate
  • fh = highest frequency of interest

If the sampling frequency is too low, aliasing of the input signal can occur (see Figure 8) so that high frequencies can appear as part of the signal in the frequency range of interest. Incorrect results are then obtained.

The solution is to use an anti-aliasing filter and the correct sampling frequency on the analogue signal, filtering out the frequency components that could cause aliasing.

Digital sine and cosine filters (Figure 7) extract the real and imaginary components of the signal.

Signal aliasing problem
Figure 8 – Signal aliasing problem

2. Relay Operating System Software

The software provided is commonly organised into a series of tasks operating in real time. An essential component is the Real Time Operating System (RTOS) which ensures that the other tasks are executed when required, in the correct priority.

Other software depends on the function of the relay, but can be generalized as follows:

  1. System services software – This is comparable with the BIOS of an ordinary PC and controls the low-level I/O for the relay such as drivers for the relay hardware and boot-up sequence.
  2. HMI interface software – This is the high level software for communicating with a user on the front panel controls or through a data link to another computer to store data such as settings or event records.
  3. Application software – This is the software that defines the protection function of the relay
  4. Auxiliary functions – Software to implement other features in the relay, often structured as a series of modules to reflect the options offered by the manufacturer.

Go back to main index ↑

3. Relay Application Software

The relevant software algorithm is then applied. Firstly the quantities of interest are determined from the information in the data samples. This is often done using a Discrete Fourier Transform (DFT) and the result is magnitude and phase information for the selected quantity. This calculation is repeated for all of the quantities of interest.

The quantities can then be compared with the relay characteristic, and a decision made in terms of the following:

  • Value above setting – start timers, etc.
  • Timer expired – action alarm/trip
  • Value returned below setting – reset timers, etc.
  • Value below setting – do nothing
  • Value still above setting – increment timer, etc.

Since the overall cycle time for the software is known, timers are generally implemented as counters.

Go back to main index ↑

Additional Features Of Numerical Relays

The digital signal processor (DSP) in a numerical relay normally can handle both the relay protection function calculations and general management of the relay such as HMI and I/O. However, if the DSP is overloaded it cannot complete the protection algorithm calculations in the required time and the protection function is slowed.

Typical functions that may be found in a numerical relay other than protection functions are described in this section. Note that not all functions are found in a particular relay. As with earlier generations of relays and according to market segmentation, manufacturers offer different versions, each with a different set of functions.

Function parameters are usually displayed on the front panel of the relay and through an external communications port.

  1. Measured Values Display
  2. VT/CT Supervision
  3. CB Control/State Indication / Condition Monitoring
  4. Disturbance Recorder (Oscillograph)
  5. Time Synchronization
  6. Programmable Logic
  7. Provision of Setting Groups
  8. Conclusions

1. Measured Values Display

This is perhaps the most obvious and simple function to implement, as it involves the least additional processor time. The values that the relay must measure to perform its protection function have already been acquired and processed.

It is therefore a simple task to display them on the front panel, or transmit them to a remote computer or HMI station.

Several extra quantities may be derived from the measured quantities, depending on the input signals available. These might include:

  • Sequence quantities (positive, negative, zero)
  • Power, reactive power and power factor
  • Energy (kWh, kVArh)
  • Max. demand in a period ( kW, kVAr; average and peak values)
  • Harmonic quantities
  • Frequency
  • Temperatures/RTD status
  • Motor start information (start time, total no. of starts/reaccelerations, total running time
  • Distance to fault

The accuracy of the measured values can only be as good as the accuracy of the transducers used such as VTs CTs, and the A/D converter. As CTs and VTs for protection functions may have a different accuracy specification to those for metering functions, such data may not be sufficiently accurate for tariff purposes.

However, it is sufficiently accurate for an operator to assess system conditions and make appropriate decisions.

Go back to main index ↑

2. VT/CT Supervision

If suitable VTs are used, supervision of the VT/CT supplies can be made available. VT supervision is made more complicated by the different conditions under which there may be no VT signal, some of which indicate VT failure and some occur because a power system fault has occurred.

If only one or two phases are lost, the VT failure algorithm can be accomplished by detecting residual voltage without the presence of zero or negative phase sequence current. For loss of all three phases, a fault in the system would be accompanied by a change in the phase currents, so absence of such a change can be taken as loss of the VT signal.

However, this technique fails when closing onto a dead but healthy line, so a level detector is also required to distinguish between current inrush due to line charging and that due to a fault.

CT supervision is carried out more easily. The general principle is the calculation of a level of negative sequence current that is inconsistent with the calculated value of negative sequence voltage.

Go back to main index ↑

3. CB Control and State Indication / Condition Monitoring

System operators normally require knowledge of the state of all circuit breakers under their control. The CB position-switch outputs can be connected to the relay digital inputs and therefore provide the indication of state through the communications bus to a remote control centre.

Circuit breakers also require periodic maintenance of their operating mechanisms and contacts to ensure they operate when required and that the fault capacity is not affected adversely. The requirement for maintenance is a function of the number of trip operations, the cumulative current broken and the type of breaker.

A numerical relay can record all of these parameters and hence be configured to send an alarm when maintenance is due.

If maintenance is not carried out within a predefined time or number of trips after maintenance is required, the CB can be arranged to trip and lockout or inhibit certain functions such as auto-reclose.

Finally, as well as tripping the CB as required under fault conditions, it can also be arranged for a digital output to be used for CB closure, so that separate CB close control circuits can be eliminated.

Go back to main index ↑

4. Disturbance Recorder (Oscillograph)

The relay memory requires a certain minimum number of cycles of measured data to be stored for correct signal processing and detection of events. The memory can easily be expanded to allow storage of a greater time period of input data, both analogue and digital, plus the state of the relay outputs.

[info]It then has the capability to act as a disturbance recorder for the circuit being monitored, so that by freezing the memory at the instant of fault detection or trip, a record of the disturbance is available for later download and analysis.

It may be inconvenient to download the record immediately, so facilities may be provided to capture and store a number of disturbances.

In industrial and small distribution networks, this may be all that is required. In transmission networks, it may be necessary to provide a single recorder to monitor several circuits simultaneously, and in this case, a separate disturbance recorder is still required.

For more information on the different types of disturbance recording, read this technical article.

Go back to main index ↑

5. Time Synchronization

Disturbance records and data relating to energy consumption requires time tagging to serve any useful purpose. Although there is an internal clock, this is of limited accuracy and use of this clock to provide time information may cause problems if the disturbance record has to be correlated with similar records from other sources to obtain a complete picture of an event.

Many numerical relays have the facility for time synchronisation from an external clock. The standard normally used is an IRIG-B or IEEE 1588 signal, which may be derived from several sources including a GPS satellite receiver.

Go back to main index ↑

6. Programmable Logic

Logic functions are well suited to implementation using microprocessors. The implementation of logic in a relay is not new, as functions such as intertripping and auto-reclose require a certain amount of logic.

However, by providing a substantial number of digital I/O and making the logic capable of being programmed using suitable off-line software, the functionality of such schemes can be enhanced or additional features provided.

For instance, an overcurrent relay at the receiving end of a transformer feeder could use the temperature inputs provided to monitor transformer winding temperature and provide alarm or trip facilities to the operator or upstream relay, eliminating the need for a separate winding temperature relay.

There may be other advantages such as different logic schemes required by different utilities that no longer need separate relay versions, or some hard-wired logic to implement, and all of these reduce the cost of manufacture.

It is also easier to customize a relay for a specific application, and eliminate other devices that would otherwise be required.

Go back to main index ↑

7. Provision of Setting Groups

Historically, electromechanical and static relays have been provided with fixed plug settings applied to the relay. Unfortunately, power systems change their topology due to operational reasons on a regular basis, such as supply from normal or emergency generation. Different configurations may require different relay settings to maintain the desired level of network protection.

Fault levels are significantly different on parts of the network that are energized under normal and emergency generation.

This problem can be overcome by the provision within the relay of several setting groups, only one of which is in use at any one time. Changeover between groups can be achieved from a remote command from the operator, or possibly through the programmable logic system.

This may obviate the need for duplicate relays to be fitted with some form of switching arrangement of the inputs and outputs depending on network configuration. Also the operator can program the relay remotely with a group of settings if required.

Go back to main index ↑


The extra facilities in numerical relays may avoid the need for other measurement and control devices to be fitted in a substation. Also numerical relays have functionality that previously required separate equipment. The protection relay no longer performs a basic protection function but is an integral and major part of a substation automation scheme.

The choice of a protection relay rather than some other device is logical as the protection relay is probably the only device that is virtually mandatory on circuits of any significant rating.

Therefore the functions previously carried out by separate devices such as bay controllers, discrete metering transducers and similar devices are now found in a protection relay. It is now possible to implement a substation automation scheme using numerical relays as the main hardware provided at bay level.

As the power of microprocessors continues to grow and pressure on operators to reduce costs continues, this trend will continue. One obvious development is the provision of RTU facilities in designated relays that act as local concentrators of information within the overall network automation scheme.

Go back to main index ↑

Numerical Relay Considerations

The introduction of numerical relays replaces some of the issues of previous generations of relays with new ones. Some of the new issues that must be addressed are as follows:

  1. Software version control
  2. Relay data management
  3. Testing and commissioning

1. Software Version Control

Numerical relays perform their functions in software. The process used for software generation is no different in principle to that for any other device using real-time software, and includes the difficulties of developing code that is error-free.

NOTE! Manufacturers must therefore pay particular attention to the methodology used for software generation and testing to ensure that as far as possible, the code contains no errors. If a manufacturer advances the functionality available in a new software version of a relay, or where problems are discovered in software after the release of a numerical relay, a field upgrade may be desirable.

This process then requires a rigorous system of software version control to keep track of:

  1. The different software versions in existence x the differences between each version
  2. The reasons for the change
  3. Relays fitted with each of the versions

With an effective version control system, manufacturers are able to advise users in the event of reported problems if the problem is a known software related problem and what remedial action is required. With the aid of suitable software held by a user, it may be possible to download the new software version instead of requiring a visit from a service engineer.

Go back to main index ↑

2. Relay Data Management

A numerical relay usually provides many more features than a relay using static or electromechanical technology. To use these features, the appropriate data must be entered into the relay’s memory and it is good practice to keep a backup copy.

The amount of data per numerical relay may be 10-100 times that of an equivalent electromechanical relay, to which must be added the possibility of user-defined logic functions.

The task of entering the data correctly into a numerical relay becomes a much more complex task than previously, which adds to the possibility of a mistake being made. Similarly, the amount of data that must be recorded is much larger, requiring larger storage.

The problems have been addressed with software to automate the preparation and download of relay setting data from a laptop PC connected to a communications port of the relay. As part of the process, the setting data can be read back from the relay and compared with the desired settings to ensure that the download has been error-free.

A copy of the setting data (including user-defined logic schemes where used) can also be stored on the computer. These can then be printed or uploaded to the user’s database facilities.

Go back to main index ↑

3. Relay Testing and Commissioning

Numerical relays perform many more functions than earlier generations of relays so testing them is more complex. Site commissioning is usually restricted to running the in-built software self-check, verifying that currents and voltages measured by the relay are correct, and exercising a subset of the protection functions.

Any problems revealed by such tests require specialist equipment to resolve so it is simpler to replace a faulty relay and send it for repair.

Quality inspection of a numerical relay printed circuit board
Figure 9 – Quality inspection of a numerical relay printed circuit board

Go back to main index ↑

Reference // Network Protection & Automation Guide by Alstom Grid

SEARCH: Articles, software & guides

Premium Membership

Premium membership gives you an access to specialized technical articles and extra premium content (electrical guides and software).
Get Premium Now ⚡

About Author


Edvard Csanyi

Electrical engineer, programmer and founder of EEP. Highly specialized for design of LV/MV switchgears and LV high power busbar trunking (<6300A) in power substations, commercial buildings and industry fascilities. Professional in AutoCAD programming. Present on

Leave a Comment

Tell us what you're thinking... we care about your opinion!

Premium Membership

Get access to premium electrical guides, technical articles and much more!