Time grading Coordination [Industrial Electrical Power Systems]

Protection design parameters for medium- and low-voltage networks

Although not appreciated by many engineers, widespread use of inverse-, definite-, minimum-time overcurrent and ground fault (IDMT OCEF) relays as the virtual sole protection on medium- and low-voltage networks requires as much detailed study and applications knowledge as does the more sophisticated protection systems used on higher voltage networks.

Introduction

Traditionally, design engineers have regarded medium- and low-voltage networks to be of lower importance from a protection view, requiring only the so called simpler type of IDMT overcurrent and ground fault relays on every circuit. In many instances, current transformer ratios were chosen mainly based on load requirements, whilst relay settings were invariably left to the commissioning engineer to determine. Most of the times, the relay settings had been chosen considering the downstream load being protected without an effort to coordinate with the upstream relays. However, experience has shown that there has been a total lack of appreciation of the fundamentals applicable to these devices. Numerous incidents have been reported where breakers have tripped in an uncoordinated manner leading to extensive network disruption causing longer down times or failed to trip causing excessive damage, extended restoration time and in some cases loss of life.

This section reviews some of the fundamental points for the design engineer to watch for in planning the application of IDMTL OCEF protection to medium-voltage switchboards and networks.

Why IDMT?

Though it may be possible to grade the relay settings based on the fault currents, it’s noted that the fault currents in a series network differs marginally when the sections are connected by cables without any major equipment like transformers in between the two ends. In such types, if networks grading the settings based on current values don’t serve the purpose. It’s required to go for time grading between successive relays in most of the networks.

To achieve selectivity and coordination by time grading two philosophies are available, namely:

1. Definite time lag (DTL), or

2. Inverse definite minimum time (IDMT). For the first option, the relays are graded using a definite time interval of approximately 0.5 s. The relay R3 at the extremity of the network is set to operate in the fastest possible time, whilst its upstream relay R2 is set 0.5 s higher. Relay operating times increase sequentially at 0.5 s intervals on each section moving back towards the source.

XX X R1 R2 R3 Fault current-- Time (s) 0.5 1.0 1.5

++++ Definite time philosophy

The problem with this philosophy is, the closer the fault to the source the higher the fault current, the slower the clearing time - exactly the opposite to what we should be trying to achieve. On the other hand, inverse curves operate faster at higher fault currents and slower at the lower fault currents, thereby offering us the features that we desire. This explains why the IDMT philosophy has become standard practice throughout many countries over the years.

R3 R2 R1 Time (s) Fault current t2, t1

++++ Inverse definite minimum time

Types of relays

Until the eighties in the last century, electromechanical disk-type relays were the standard choice, the most popular being GEC's type CDG and the TJM manufactured by Reyrolle.

They both follow the IEC specification for the normal inverse curve as highlighted, with an acceptable error margins as identified in the graph.

++++ IEC/IEEE tolerance limits

Being of the moving disk principle, the disk has a tendency to 'overshoot' before resetting after the fault current is removed by a downstream breaker. This phenomenon has to be considered, together with the tolerance on the tripping characteristic coupled with the breaker clearing time when determining the optimum time grading interval of 0.4s as developed. However, on the modern electronic digital versions there is no overshoot to worry. In addition, they offer better tolerance over the whole curve - better than 5% is claimed - so the combination of these two factors means the time-grading interval can be reduced to 0.3s. Another point often overlooked in the use of electromechanical relays is that the burden of the relay varies inversely with the plug setting. The lower the plug taps setting, the higher the burden.

This had been illustrated in the previous section where it had been noticed that for a 1A relay the range varies from 0.75 ohm on the 200% tap to 300 ohm on the 10% tap. Similarly for the 5A relay the range varies from 0.03 ohm on the 200% tap to 12 ohm on the 10% tap.

The choice of plug tap could therefore have a significant effect on the performance of the current transformers to which the relay is connected.

++++ Typical disk-type relays

This shortcoming has been addressed on modern electronic relays, the burden remaining constant over the whole setting range and at a very low value, typically 0.02 ohm as seen in the previous section, without any major implication on the settings adopted.

Network application

When deciding to apply IDMT relays to a network, a number of important points have to be considered.

Firstly, it must be appreciated that IDMT relays cannot be considered in isolation. They have to be set to coordinate with both upstream and downstream relays. Their very purpose and being is to form part of an integrated whole system. Therefore, whoever specifies this type of relay should also provide the settings and coordination curves as part of the design package to show that he knew what he was doing when selecting their use. This very important task should not be left to others and once set, the settings must not be tampered (even by the operating staff ) as otherwise coordination is lost.

++++ Minimum grading intervals

Overcurrent grading:

When assessing the feasibility of applying such protection, one must be aware of certain constraints that will be applied by the supply authority at the one end and the coordination requirements of the low-voltage network at the other. These factors can often place severe limitations on the number of grading steps that can be achieved. It will be seen that one could soon run out of time on overcurrent relays at the downstream side and it may be impossible to provide the settings for the downstream end relays.

++++ Typical microprocessor relays, Supplier, Consumer

++++ Overcurrent-time grading intervals --0.4 s Interval; 0.3s Interval

++++ Influence on network design (short and fat)

++++ Influence on network design (long and thin)

Using modern electronic relays does help in that with a 0.3 s time interval a couple of extra grading steps can be gained. The message here is therefore very clear. When designing medium-voltage network, one must aim for the minimum number of grading levels. In other words, medium-voltage networks should be designed 'short and fat' rather than 'long and thin', (b). If this recommendation is not followed, then one may be faced with running a network radially by opening a ring at a specific point in order to achieve some moderately acceptable form of grading. The relay settings would thus be dictating at which point the ring must be open. This is a bad design practice as protection relays should never place any limitations on which way the network should be operated, as maximum flexibility is essential at all times. Furthermore, running radially means continuity of supply is lost under fault conditions and coordination would be lost when the network is re-arranged to restore supply from an alternative source.

++++ Impact on system configuration

++++ Ground fault -- * LV ground fault means very small hv overcurrent

* Ground fault level tends to remain constant

Ground fault grading

Generally, MV systems are impedance grounded at its neutral end, which limits the ground fault current to around 300A. ++++ the current distribution for an MV ground fault from which it will be seen that only a small overcurrent flows on the HV side of the supply transformer under a fault. This is of no embarrassment to the supply authority so they often allow greater clearance times for ground faults on the consumer's network. Times of 3.0-3.6 s are typical for the supply authority's back-up protection.

This allows for a number of grading steps to be achieved for the ground fault protection into the consumer's network.

However, a point that is not often appreciated is that the NEC is the dominant impedance in the ground fault loop. This means that an ground fault anywhere on the MV network is controlled to a virtual constant level of current by the NEC. If the ground fault current is constant, then the IDMT EF relays behave as definite time relays, operating longer for a fault closer to the source. An ground fault near to the source can therefore easily develop into high-current phase fault before the ground fault relay has timed out leading to enormous damage.

Transformer protection:

IDMT relays have often been used as the main HV protection on distribution transformers, without due regard for certain limitations.

It’s often very difficult, in fact impossible, to set an HV IDMT relay to detect an ground fault on the LV winding of a transformer. As will be seen the equivalent HV current for an ground fault on the LV winding (especially towards the neutral end) can be below the primary full load current. If the LV winding is grounded via a resistor the ground fault current won’t exceed 57% of the full load rating of the transformer. The transformer should therefore be fitted with an additional protection such as Buchholz or restricted ground fault (REF) on the LV side to cover this condition.

It has been common practice not to fit Buchholz relays to small transformers because it’s too expensive relative to the cost of the transformer. However, the size of the transformer is not that important but its strategic location in the network counts. Its loss may have a major impact on production downtime, etc., irrespective of its size, so it’s worth protecting it properly and the Buchholz alarm does give early warning of impending trouble.

++++ HV overcurrent on transformers -- Fault position from neutral

++++ Fault current due to grounding resistor Another poor design practice is for the HV IDMT relay only to trip the HV breaker. For an HV winding fault the breaker is tripped but the fault can continue to be fed via the low-voltage side, the back-feed coming from the adjacent transformer(s) where the LV protection is set high to coordinate with downstream requirements. Transformer protection should always trip both HV and LV circuit breakers wherever possible as shown.

++++ Protection should trip HV and LV

This is very much important when a low-voltage bus is connected from more than one source. In some cases, it’s a common practice to ensure that the secondary breaker is tripped whenever the primary breaker is tripped/open (intertrip) and to introduce interlock to avoid secondary breaker being closed without primary breaker in closed position.

There is also a general misconception that the HV IDMT overcurrent relay is there and to be set to cover overloading of the transformer. It’s not true since the overcurrent relay acts more as a protection against faults and not as overload protection. If the thermal characteristic of a transformer is plotted against the normal inverse IDMT curve (which is on log-log graph scale) it will be seen as a straight line crossing the IDMT characteristic at some point. It may be possible to select the settings such that for small overloads, the relay will trip before the transformer heats up to its limit. However, for sudden heavy overloads the transformer will cook before the relay trips. The normal inverse IDMT relay is therefore NOT suitable for overload duty - it’s a fault protection. If overload protection is desired then a relay with a suitable thermal characteristic should be applied or alternatively select the inverse characteristic in the IDMT range, the time being approximately inversely proportional to the square of the current. This characteristic is much closer to the thermal characteristic of a transformer and any fuses downstream in the LV network.

Time (s) Current IDMT extremely inverse Transformer thermal characteristic AB IDMT normal inverse Light overload relay trips before transformer cooks A=

Heavy overload transformer cooks before relay trips B=

++++ IDMT normal inverse not for overload.

However, why trip the transformer out for overload - why not just remove the load by tripping the LV breaker? The same current that flows through the transformer flows through the LV breaker and nearly all have thermal characteristics as an integral part of their protection. This approach would certainly save operators having to test and check out a transformer before returning it to service, thereby reducing downtime.

Current transformers

There are numerous installations where the performance of the current transformers has been overlooked or misunderstood.

The most widely mistaken error has undoubtedly been the choice of CT ratio which was invariably selected on the basis of full-load current without due regard to performance under fault conditions. Despite its name, the performance parameter that we are most interested in is the current transformer's secondary voltage. In other words, the voltage developed across the CT secondary shall be sufficient to drive the transformed current through its own internal impedance plus the impedance of the relay and any other equipment to which it’s connected. It’s necessary to fix the typical form of a magnetization curve for the CT. The knee-point voltage and the CT's internal resistance should be specified/checked to ensure that they are adequate for the application.

A classical example of how-not-to-do-it was seen recently on a feeder to a mini-sub from a main 11 kV switchboard whose fault level was approximately 9000A.

CT ratio: 25/5 Burden: 5 VA Accuracy: 10P 10 Overcurrent relay: CDG set at 50%.

In the above example, for a 9000A fault on the feeder cable, the CT secondary current would be 1800 A. The burden of the relay at the chosen setting is 0.48 ?, therefore the secondary voltage required to drive 1800 A through an impedance of 0.48 ohm = 1800 × 0.48 = 864 V. With five turns on the CT the best that can be expected is 1 V per turn which equals 5 V.

IMDT relay ATH =0.2 A =0.04 RL =0.01 V IS = 400 A RCT =0.01 50% TAP = 0.48? 100/5 10P10 8000A 800 kVA IFL =42A 200% TAP = 0.03? Primary As = Secondary As 100 A × 1 turn = 5A × 20 turns ++++ CT performance important Mini-sub IDMT 9kA 25/5 5VA 10P10 CDG Set at 50%.

++++ Typical selection of CT.

It will now be evident why the magnetization curve and internal resistance of the CT are so important, especially when electromechanical relays are used for protection. With such a low ratio, this CT would undoubtedly be of the wound primary type in order to get the required number of amp-turns to magnetize the core. Unless the CT manufacturer was told, the high fault current would cause this CT to burst due to the high thermal and magnetic stresses set up in the primary winding under fault conditions which are a function of the fault current squared. CT ratios should therefore be chosen based on fault current NOT load current.

In such applications, the use of 1A secondary is recommended as there would be one-fifth of the current and five times the voltage available to drive this smaller current around its connected load. The accuracy stated in the above example of 10P10 means that the CT will remain 10% accurate up to 10 times its rating (i.e. up to 250 A) -- not much use for a fault current of 9000 A. Ideally, the chosen ratio should have been 900/1 10P10 or 450/1 10P20. The correct technique in choosing a CT ratio is to calculate the fault current and divide this by 10 for a 10P10 accuracy rating or by 20 for a 10P20 specification. Accuracy is important, especially in tight grading applications as highlighted above.

It’s also a common practice by some designers to add an ammeter into the protection circuit. This is invariably done for reasons of economy to save fitting another set of CTs.

However, this is not a good practice because:

1. It adds more burden into the CT circuit -- typically 1 VA for a normal ammeter and 5 VA for a thermal maximum demand ammeter -- thereby giving the CT more work to do, thereby pushing into saturation.

2. Ammeters are designed to operate under healthy conditions over the range 0-1.2 times full load. Ideally, they should be connected to metering cores, which saturate at this upper level to protect the instrument under fault conditions.

3. Protection cores on the other hand are designed to operate under fault conditions at 10 to 20 times full-load current. Connecting ammeters to protection cores therefore subjects the instruments to enormous shocks under fault conditions, which initially affects the accuracy but ultimately destroys the bearings, and mechanisms, etc.

It’s evident that this common wrongful practice of adding meters onto protection cores is the main influencing factor in deciding the CT ratio. A classic case of the tail wagging the dog!

++++ indicates the CT ratios and the settings adopted in a typical network comprising of various transformers. It can be noted that the CT knee-point voltages as per the actual fault current figures are quite different from the actual knee-point voltage of the available CTs (refer lines marked '?'). It’s a very clear case, wherein the wrongful selection of current transformers could lead to relays not operating at fault conditions.

This shall be avoided, which can be done by doing proper calculations before deciding the component details.

Take the settings for a transformer in line 3. The fault current is around 6200A and with a CT ratio of 100/5 A, the relay, current under fault will be about 310 A. This requires a knee-point voltage of not less than 164.5 for the provided CT but the actual knee-point is only around 46 V. The relay is not going to serve any purpose under fault conditions. Just consider the CT ratio of 100/1 A. The relay current under fault will reduce by one-fifth, and the knee-point voltage required will be only around 33 V and the available CT will do its job without any problem. This kind of analysis is necessary while choosing CT ratio and knee-point voltages while designing a network, which requires coordination for relay grading.

Importance of settings and coordination curves

One can have the finest protection in the world, correctly designed, installed, commissioned and maintained but if it’s not set correctly then it’s not of much use.

Careful attention should therefore be paid to the settings of IDMT relays in particular, as they have to coordinate with upstream and downstream relays. To repeat what has been stated previously, the person who selects and specifies this type of relay should therefore also provide the settings and coordination curves to prove that the relay can fit correctly into the new or existing network. Many instances have arisen where it has not been possible to achieve any sort of adequate settings (as has been highlighted in the examples above) thereby proving the designer did not know what he was doing, or misunderstood the fundamentals.

In setting the IDMT relay it must be appreciated that moving the plug bridge moves the curve left or right in the horizontal direction and selects the current pick-up value.

Adjusting the time multiplier dial moves the curve up or down in the vertical direction to select the time of operation.

++++ Typical case of CT selection in a large network: Multiplier dial; Plug setting time

++++ Effect of settings.

It’s therefore possible to achieve the same setting using two different combinations of plug setting and time multiplier. Notice how the curves cross. Even with a time interval between them, it’s still possible to choose plug/time dial settings such that the curves cross. This means that for low fault currents relay A operates before relay B but for high fault currents relay B operates before relay A. Coordination is therefore lost. It’s vitally important that after selecting settings for the relays, the coordination curves are drawn to ensure that they don’t cross and that they stack up nicely on top of one another.

Same setting using different combinations of plug and time multiples.

++++ Curves must not cross; Min. Max.

++++ Ideal coordination of setting curves.

Two basic rules are that they must pick-up for the lowest fault level (minimum plant) and must coordinate for the highest fault level (maximum plant). One final word on a point not often appreciated. The electrical network is in fact a living thing. It grows and changes over time. Generation is added, load centers develop, old plant's decommissioned, new plant's extended and so on. Fault levels therefore change and invariably increase. IDMT relay settings should therefore be reviewed on a regular basis, especially if there are extensions or changes planned. The opportunity should be taken to up-grade the protection at the same time. The old electromechanical disk relays have done a wonderful job over the last century and can still continue to do so in certain applications provided one is aware of their limitations. However, we are now starting to ask more than they are capable of delivering. Nothing wrong with the relay -- it's just the application. So one should take advantage of the additional facilities offered by the new range of numerical relay equivalents.

Conclusion

To engineers planning the protection for a medium-- to low-voltage network and wishing to adopt the widespread use of the IDMT OCEF relay, the above can be summarized as follows:

• Design networks with the minimum number of grading levels possible.

• Choose CT ratios based on fault current -- not load current.

• Consider using 1A secondary.

• Check CT magnetization curves for knee-point voltage and internal resistance.

• Connect ammeters, etc. onto own metering cores.

• Provide relay settings and coordination curves as part of the design package.

• Be careful when choosing relay plug tap setting on electromechanical relays.

The lower the tap, the higher the burden.

• Relays should not pick-up for healthy conditions such as permissible transient overloads, starting surges and reconnection of loads, which have remained connected after a prolonged outage.

• Care should also be taken that the redistribution of load current after tripping does not cause relays on healthy circuits to pick-up and trip.

• HV IDMT relays on transformers should trip both HV and LV breakers.

• Normal inverse curves should not be selected for overload protection. Rather use the inverse characteristic for this duty.

• Take advantage of the additional features offered by the modern electronic relays, e.g. fixed very low burden, integral high-set, breaker fail and busbar blocking protections, event memory, etc. However, remember, one has to do the same calculation exercises for settings and draw coordination curves whether the relays are of the electronic or electromechanical design.

• Finally, if the switchgear suppliers also manufacture relays, don’t expect them to do the protection application settings free of charge as part of the service.

If this is required, specify this as a separate cost item in the specification.

Many problems down the line can be avoided and the performance, efficiency and safety of the plant improved if a protection engineer is included in the design team, if not full time, but at least to do an audit on the proposals.

Finally, remember -- whilst IDMT relays are the most well known and the cheapest, they are in fact the most difficult relays to set.

Sensitive ground fault protection

A number of instances arise where the load current demands high-ratio line current transformers, and the neutral current has been limited to a low value by the use of a remote high-impedance neutral grounding device for safety reasons.

In this situation, the use of a conventional IDMTL ground fault relay with a minimum setting of 10% would be unable to detect an ground fault condition. For such an application, sensitive ground fault protection should be considered. In addition, on overhead rural distribution systems it’s possible for high-resistance ground faults to occur, especially if a conductor breaks and falls on very dry ground having say high silica content. Instances have been recorded where the initial rush of fault current has caused the silica to form a glass envelope around the end of the broken conductor; so a live conductor then remained undetected following the auto-reclose shot.

++++ Ground fault current limited by NER: 600/5 10 A R E/F

Besides this type of unusual incident, it’s not uncommon for broken conductors to fall on dry ground, and this presents a hazard to human life and livestock if left undetected for any length of time, particularly if the line does not have ground wires installed. It’s therefore necessary to apply sensitive ground fault relays to cover these and other demanding situations.

Another factor is the possible predominance of a third harmonic component in the residual current even under quiescent conditions. Third harmonics appear as zero sequence currents and could cause mal-operation of the relay. It’s therefore wise to select a design of relay, which has been tuned to reject currents of this frequency. Also, ensure that the response decreases at higher frequencies to render the relay immune to harmonic resonant conditions, which may vary according to the power system configuration. There are a number of sensitive ground fault relays on the market, essentially of the definite-time variety, having pick-up setting ranges typically of the order of 0.4-40% of 5A. One model offers a useful digital read-out of the standing neutral current when interrogated via a push-button. If this is recorded on a regular basis, trends can be established which would assist in determining the deterioration of the line insulators and help to plan preventative maintenance programs.

Time setting ranges vary from 0.1 to 99sec. Adjustable time delays ensure stability during switching and other transient disturbances and allow for adequate grading with other protection systems.

On rural networks, it’s as well necessary to ensure that the relay contacts are self-reset (i.e. they don’t latch) so that auto-reclosing can take place.