TECHNICAL PAPER

Loss of Mains Protection

Theory and Application

Contents 1 Preface 2 Basics

2.1 Purpose and meaning of mains decoupling 2.2 Operational consequences 2.2.1 From the private power producer point of view 2.2.2 From the Electrical Supply Authoritys point of view 2.3 Reasons for loss of mains protection 2.3.1 Manual disconnection 2.3.2 Short-term mains interruptions with automatic reclosing 2.3.3 Single-phase failure 2.3.4 Voltage recovery 2.4 Dynamics of mains disconnection 2.4.1 Mains disconnection without preceding voltage change 2.4.2 Mains disconnection after preceding voltage change

3 Decoupling Criteria
3.1 Supervision of single quantities 3.1.1 Voltage (U<, U>) 3.1.2 Voltage unbalance 3.1.3 Reverse power (PINV) 3.1.4 Frequency (f< , f>) 3.1.5 Rate of change of frequency (df / dt ) 3.1.6 Vector surge 3.1.7 Load change (P) 3.2 Combined criteria 3.2.1 Overcurrent and undervoltage (I>& U<) 3.2.2 Directional overcurrent (I> & U<) 3.2.3 Underfrequency and undervoltage (f< & U<)

4 Function of Relays Available

4.1 Rate of change of frequency 4.1.1 Mean value detection 4.1.2 Quasi-instantaneous value 4.2 Vector surge measurement

5 Example for Relay Applications

5.1 Embedded generator with possibily of island operation 5.2 Combined heat and power plants 5.3 Long closed-loop network 5.4 Wind power station 5.5 Embedded generators as UPS

6 Summary

Mains decoupling

Preface

Due to the ecopolitical trend for better utilization of primary energy and process heat, the mains parallel operation of smaller and medium sized power generating systems gains more and more in importance. Besides industrial small-scale power plants there is also a growing number of Combined Heat and Power Plants (CHP) or private power generating systems which are partly using as well alternative energy sources. In some cases, though the mains parallel operation of several generators can cause some problems and especially when failures occur in the public grid, the operator has normally to ensure that the power generating system is decoupled automatically from the faulty remaining grid. After disconnection of the inter-tie circuit breaker, however, the own power supply ought to be maintained as an autarkic island. This paper now throws light on criteria which allow to identify mains failure conditions which must lead to disconnection of embedded generators. Measuring procedures as well as their suitability and limits are explained. The main focus lies on the measuring procedures of vector surge detection and evaluation of the rate of change of frequency (df/dt or ROCOF). By stating specific fringe conditions this paper is meant to be a help for chosing the right measuring procedure and to set parameters of the relevant protection device suitably.

Mains decoupling

Basics

2.2.1 From the private power producer point of view Normally the private power producer endeavours to supply his section of the mains independently and without interruptions if the power requirement does allow this. Generally there is no problem to detect faults in the own mains section, because they can be identified selectively and handled. This, however, is different with failures occurring in the public grid and if these effect the own system. For identifying the fault the private power producer has only the quantities being measured at the utility connection point. The primary quantities are: current I voltage U frequency f The secondary quantities are: active power P reactive power Q apparent power S The diverted quantities are: load change dP/dt rate of change of frequency df/dt vector surge On switching operations outside the own section, the private power producer has no influence whatsoever. There is only one thing he can do to eleminate an external fault and that is to open the inter-tie C.B. This measure not only protects the own consumers from a voltage breakdown but also the generator from the following problems : At short circuit: overload thermal destruction At asynchronous reclosing: thermal overload destruction of winding heads breakage of the coupling breakage of the driving shaft 2.2.2 From the Electrical Supply Authoritys point of view The operator of the public grid, too is very concerned to keep the supply interruption times at very low levels indeed. But to have a maximum of operational possibilities at mains failures, private power producers must not be a restrictive factor here. For this reason very often the Electrical Supply Authorities combine their system connection permissions with special compulsory requirements. These requirements may include the directive to use specific protection measures, exclude certain operational methods or perhaps give periods of time where energy exports are allowed only.
Mains decoupling

2.1 Purpose and meaning of mains decoupling

For all those parts of the mains which are used for continuous feeding of customers, supply failures are always a disruptive element and sometimes they even entail risks and involve considerable costs. Therefore it is an understandable objective of mains operators to keep number and duration of interruptions as low as possible and to be able to switch off the faulty section selectively. Hence it is important that at an early stage, i.e. during mains and protection studies, precautionary measures are taken with regard to detection of probable and possible failures and how to counteract these by suitable protection systems. For this the following questions should be answered: Which mains failures are possible? Which switching operations in the mains can be dangerous for other sections? Which switching operations and faulty conditions have to be considered for protections measures? How can the individual failure be identified? What has to be done in case a failure occurs taking into account the prevailing state of the art? What are the protective measures required by the local Electrical Supply Authority? What has priority in case of failure: the system protection or supply continuity? What switching operations are required in case of failure? What protection measures are additionally necessary?

2.2 Operational consequences

The multiplicity of possible proceeding variants become very clear by the questions raised in the preceding paragraph. If the different mains sections are not monitored by a common control system, a mains failure has to be identified by all operators of power generating systems at the same time and counteracted accordingly by protective measures. Especially where protective measures are concerned, the private power producer may contractually be bound to specific requirements set up by the local Electrical Supply Authority and so could be faced with a clash of interests. The views of Electrical Supply Authorities in the different countries differ generally to quite an extent. The measures to be taken in case of failure depend, of course, on the significance of the private system within the public grid. The kind of operational methods used have also an influence on the way of approach to rectify the fault.

2.3 Reasons for loss of mains protection

In this paragraph a number of system faults and switching actions are given which must result in decoupling of the private supply system to avoid the risk of endangering the own system or the own generator. Decoupling means opening of the closed inter-tie C.B. where the private power system and public grid are connected. In some cases this can even be the generator C.B.

The generator has not enough power reserve to feed the island Here too the island loses its synchonism with the public grid, without the possibility for direct reconnection. In addition the generator is overloaded which may even be of short circuit type nature. Uncontrolled decrease of the isolated system frequency and the mains voltage are the result. If the generator can at least supply the own system, the problems for the island may be solved in by mains decoupling. 2.3.2 Short-term mains interruptions with automatic reclosing According to statistics many short circuits in overhead lines could be cleared by disconnection of the faulty phase so that after reconnection the line is in working order again without the need of any other measures. An electric arc which was ignited by an alient element in the insulation media keeps burning as long as it receives the necessary energy to ionize the insulation media. The failure source (e.g. lightning stroke, birds, branch) very often does not exist very long and in such a case a short-term disconnection of the voltage is sufficient in order to initiate recombining of the ionize insulation gas (mostly air) in the arc channel. Then the "insulation gas" fulfills its initial job again and is not conductive any more; the arc extinguishes. When the voltage is reconnected after about 200 - 400 ms, the line is faultless and is operable again. For private power producers the automatic reclosing procedure entails some problems. After disconnection of the voltage, the isolated system and remaining grid can very quickly lose their synchronism and when reconnected both sections may meet out of phase. The adverse effects have already been listed in para. 2.2.1. If there are C.B.s in the system around the private power producer which are povided with an automatic reclosing facility, it is absolutely vital to take measures for a fast decoupling of the generator from the mains. For this a protection relay has to identify the nature of the fault quickly so that sufficient time is left for reaction of the command chain to the C.B. before the dead time of the automatic reclosing system elapses. Proceedings can be entirely different if many small private power producers have a considerable share in the total power supply of a region. In such a case it is important that not all supplier disconnect themselves from the intrasystem immediately because if they would, the power supply of the whole region would be at risk and selective disconnection impossible. Here other protection methods than fast decoupling are needed (e.g. time overcurrent differentiation without immediate tripping or auto reclosing facility with synchronization control).

2.3.1 Manual disconnection If the Electrical Supply Authority wants to disconnect a specific section of the mains, e.g. for checking purposes or earth fault detection, all private power supplier connected must decouple their system so that the specific section becomes really free of potential. If disconnection is planned, this can be coordinated with the private power supplier, but it is also possible that disconnections may happen by mistake or caused by a technical fault. If in such a case the switching party is not informed of the feeding action of private power suppliers, then manual reconnection of that section is extremely dangerous. Normally it is unlikely that synchronization at the opened isolating point in the public grid is possible. The isolated mains sections will soon lose their synchronism and it is then not possible to close the opened C.B. without risk. This section must be reconnected at a synchronizable isolating point and this very often is in the system of the private power supplier. If the Electrical Supply Authority disconnects the section where a private power supply system lies, the following two techn. situations with consequences for the isolated section are possible: The generator has enough power reserve to feed the island. The disconnected mains section is not potential free - as intended. The island loses its synchronism with the public grid. Disconnection can normally not be reverted at the same C.B. if that C.B. is not resynchronizable. This condition can exist for a long time without being noticed because there are no direct effects. There is a danger of critical voltage and frequency conditions and when damage is inflicted on a third party, the causative private power producer can be faced with claims for damages.

Mains decoupling

2.3.3 Single-phase failure A single-phase conductor failure can be the aftereffect of a blown fuse or a broken conductor. However, the generator now only being connected with the mains via two phases, does not lose its synchronous connection. If there is no other reason (e.g. short circuit or inadmissible unbalanced load), fast mains decoupling is not advisable, although the specific mains section must be disconnected the latest when the fault is being rectified.

Example 2: Our second example refers actually to the same grid but with the difference that the fault has occured in an adjacent outgoing near by the transformer station.
Voltage
1 00

% Un

Trans form er S ubs ta tion Junc tion Line

2.3.4 Voltage recovery This chapter explains how an out of phase voltage recovery endangers the mains parallel running generator. Example 1: A connection line is provided between the private power supplier and the transformer station. Along this line a consumer with short circuit protection is connected.
Voltage
100 % Un

Fig. 2-2: Short circuit in an adjacent outgoing

Here, too a voltage valley develops near the fault and there generator and mains voltage may drift apart. If the faulty outgoing is disconnected, it can happen that voltages collide in asynchronous conditions. The farther the distance between location of the short circuit and the station, the more unlikely drifting apart of the voltages becomes because between transformer and generator the voltage does not drop so extremely.

Trans form er S ubs tation Junc tion Line O ver C urrent Protec tion (e.g. F us e) Load

2.4 Dynamics of mains disconnection

As it can be seen from the preceding chapters, the mains can be decoupled either without preceding voltage changes (e.g. manual switching off, overload tripping) o r

Fig. 2-1: Short circuit in a connection line:

In the example (Fig. 2-1) a polyphase short circuit has happened at the point indicated. Between occurence of the fault and the time the protection becomes effective, the voltage on the connection line has about the characteristic as shown. A voltage minimum exists at the point nearest to the short circuit and because of the low voltage content, the generator link to the remaining grid is of lowest intensity. Compared to a driving shaft, which becomes more and more conical tapered towards the middle and then breaks at the thinnest point, here the generator voltage can turn away from the mains voltage. Shortly after the fault has occured, the overcurrent protection will become effective and disconnect the faulty outgoing. At this moment the voltage on the line recovers. Mains voltage and generator voltage are now in direct opposite position. If the two voltage were not synchronous any more in the instant of disconnection, transient processes come into effect, similar to those during asynchronous connection. By this the mains and generator are extremely stressed. This example shows clearly that an asynchronous voltage restoration is absolutely possible after disconnection, even without automatic reclosing.

after a voltage change (short circuit with or without automatic reclosing, conductor disruption)

Mains decoupling

2.4.1 Mains disconnection without preceding voltage change A synchronous generator, which had exchanged energy with the Electrical Supply Authority via a connection line, changes to isolated operation after a C.B. had opened (see Fig. 2-3). In such cases it is possible that a residual load remains at the generator side and this load the generator carries on supplying.

G
L o ad

Ju n ctio n L in e

G rid

For a synchronous motor generally the same applies as to a synchronous generators, the significant difference is that the motor is not driven but just can use its flywheel for feeding energy into the mains. The speed decreases instantly; the voltage behaviour is conditional on the excitation status defined before mains decoupling. In case the energy of a private power generating system is fed into the mains via an inverter, the active power and reactive power output drops very quickly to zero. Only for a short time after mains decoupling the voltage of the isolated system is backed slightly by asynchronous motors connected to the grid. 2.4.2 Mains disconnection after preceding voltage change Voltage changes can have different reasons (e.g. short circuit with or without automatic reclosing). Since it is intended that the mains is decoupled directly after a short circuit -and this applies especially to systems where the connection lines are provided with auto reclosing facilities- the mains transient for the generator and the driving genset is a short circuit with following quick clearance (see Fig. 2-4).
Co up ling Switch AR G rid S h o rt C irc u it

Fig. 2-3: Forming of an isolated system after switching off manually

Conditional on the active and reactive power component in the isolated system, the following dynamic behaviour can be predicted after first approximation: When the active power load increases during forming of the isolated system, the generator speed decreases and so that of the isolated system. If less active power is fed to the isolated system, speed and frequency values increase accordingly. When the reactive power load increases during forming of the isolated system, the generator terminal voltage decreases. If less reactive power is fed, the terminal voltage increases accordingly. When loading of the generator increases through active power and reactive power during forming of the isolated system, the voltage vector of the generator and by this the residual grid drop back immediately within a few cycles. When the generator is deloaded, the voltage vector advances accordingly. Even after the "jump" the voltage vector keeps changing its position according to this frequency behaviour: If the load is further increased, the vector drops back even further and advances more and more at deloading. Those thumb rules are valid for the transient time range 10-300 ms after the mains failure. After this time the speed / frequency or voltage of the driving genset and the generator are controlled by their regulators. N o t e: As to the first moments of the explained transients, the afore said normally applies to asynchronous generators and big asynchronous motors as well because their "excitation energy" derives at first from the rotor current (for slip reasons).

G
L o ad

J u nc tio n L in e

Fig. 2-4: Forming of an isolated system by a short circuit and automatic reclosing

During the first phase of the mains failure the voltage at the generator terminals collapses according to the impedance ratio. Due to the high-inductive short circuit angle of the connection line, the voltage at the generator terminals become highly inductive as well. This now results in advancing of the (reduced) voltage vector. Because the active power at the generator decreases at the same time, the speed increases (slightly) and the voltage vector advances even further. By opening the circuit breaker the mains failure enters into the second phase where the faulty section of the line is switched off from the generator and from the remaining load. Only now the voltage recovers and on the whole the dynamic force is similar to that described before, conditional on the active and reactive power component in the isolated system. Also when the mains is as quickly decoupled as in this case, it can be assumed that the dynamic of mains decoupling progresses without influence until speed / frequency or voltage are controlled by the regulators.

Mains decoupling

Decoupling Criteria
Overvoltage detection too is an important criterion for decoupling purposes. If, for instance, a generator becomes suddenly deloaded by decoupling and through isolated operation, the voltage at the generator terminals can rise very quickly to values which can be a risk for some of the consumers. As part of the generator protection an overvoltage protection is adviseable, but it might be too slow for recognizing mains decoupling.

In this chapter measuring procedures are explained which identify the specified mains failure. Whether the stated procedures are applicable, however, depends in the individual case on the operational conditions of the power generating system and the grid.

3.1 Supervision of single quantities

In this section procedures are described which use single measuring quantities as decoupling criteria.

3.1.2 Voltage unbalance A voltage unbalance relay monitors the symmetry of voltage vectors in three-phase systems. Resolving of a vector system into symmetrical components Positive phase-sequence system Negative-phase sequence system Zero-phase sequence system is a method commonly used. When a short circuit occurs in one phase or two phases, the voltage vectors of the faulty phases deviate from their normal values. This effect is reflected in rise of the amplitude of the negativephase sequence system and in voltage drop of the positive phase-sequence system. The amplitude of the zero-phase sequence system U0 is a quantity for displacement of the voltage vector star point from its normal position. The zerophase sequence system gives information e.g. about earth faults in isolated and compensated systems. Compared with a mere RMS value measuring, the method of symmetry supervision has the following advantages: A single-phase disruption (fuse trip, conductor flaw) in a nearly no load mains section might be not recognizeable by an undervoltage element. It is quite possible that transformers, transducers or parallel operating generators reproduce part of the missing phase and so voltage is still applied at the faulty phase, but the phase angle of this voltage vector does not tally with the normal one. The relevant voltage vector system is asymmetrical which can be seen from the value of U2 . U1 U2 U0

3.1.1 Voltage (U<, U>) Here the term "voltage measuring" is specifically understood as RMS value supervision. For the intended purpose the voltage is very suitable to be used as measuring quantity because in the ideal case it is nearly identical at all measuring points in the mains. Aided by the voltage it is also possible to supervise a widly branched grid right into every feeder. Measured at the decoupling point, overand undervoltages are always a sign that decoupling is necessary because of the risk to own and other consumers connected. Loaded motors, for instance, need a higher current when the mains voltage is reduced and this can result in thermal overload or even stalling. In a system with self-generated power the success of decoupling in case of undervoltage depends on the capacity the generator or generators operating in parallel have. If the generator is able to cover nearly all the own power demand, it is likely that the RMS value of the mains voltage will hardly differ during the dead-time of an auto reclosing cycle, if the fault has occured in a different branch of the mains. At this moment especially asynchronous motors will function as generator and so back-up the mains voltage for a short time. In systems without self-generated power, however, decoupling as a result of undervoltage is only adviseable when restarting of machines after voltage recovery has to be prevented because of the risks involved. If the switching point of the voltage relay (U<) is adjusted such that it is acting too fast, this may result in trips due to operation related voltage variations. Start of a motor can also cause a shortterm voltage drop which has to be permissible. At the other hand, if undervoltage tripping is adjusted such that it is acting too slow, or the time of the switching delay is too long, then fast decoupling is slowed down too much. This means that undervoltage detection as a fast mains decoupling criterion is not always suitable.

Mains decoupling

3.1.3 Reverse power (PINV) With this protection method the direction of the electrical energy flow at the mains coupling point is monitored. The precondition is that in normal operation energy import only is permitted. Where such restriction is acceptable, this protection method means a secure and quick indication for necessary mains decoupling. If at a failure the own generator supplies external consumers, the coupling switch has to be opened. For plants with constant energy export or with changing energy flow direction, this method is not suitable.

values can be considered and by this the tendency of frequency change can be analysed. There are two different requirements on a frequency trend analysis: Fast tripping Security against nuicance trippings To achieve more security against nuicance trippings, the frequency has to be observed for some time (several measuring intervals) by which the tripping time is delayed. Switching operations in the mains, however, may displace some zero passages and so fake two contrary frequency changes with the effect that for instance, the first cycle is measured too short and the following one measured too long; Increasing frequency followed by a decreasing one is detected by relays. As precaution against trippings, the relay therefore has to evaluate the direction of the frequency change and its continuity. This requirement is met by digital protection relays, although extensive algorithms are necessary to eliminate the noise of measuring faults in frequency measuring values. A change of, for instance, 1 Hz/s causes the cycle duration to alter just about 4 s per cycle. As comparison : The normal cycle duration at 50 Hz is 20 ms. This difference requires a high timing accuracy of the relevant relay so that frequency changes can be recognized precise enough. As to the measuring side of this method these were the essential facts. A different problem is how to define the correct tripping value. When considering the quickest possible tripping time at maximal security against trip conditions, the architecture of the mains has to be taken into due account. Not only the power supplier but the consumers as well have an influence on the extent of frequency changes. Influencing parameters are: Kind of drive (inertia) Kind of consumers (own ones and those of the public grid) Kind of switching operations In the following Fig. 3-1 a typical frequency behaviour after shut-down of the mains is shown. Prior to shut-down the isolated system was supplied by the public grid partially. At first the generator speed decreases constantly because the generator has additionally to supply the mains consumers. Later on the engine governor intervenes and tries to correct the speed dip.

3.1.4 Frequency (f< , f>) During mains parallel operation normally the load of a generator prime mover is power regulated. This means that the driving torque is adjusted to the required electrical energy output. The speed of the genset does not change because it is defined by the interconnected mains. If a power generating system changes into isolated operation, its regulating system should be informed about the new status without delay. In case the regulating system does not receive this information. the speed will be changed by the resulting load jump. By the increasing load the generator speed decreases and so does the frequency of the isolated system. Analogously to this the frequency rises when the generator is deloaded. Compared to steam turbines, the reaction of diesel generators is slower. This shows that failure of the mains supply can be recognized by frequency supervision. By using additional frequency supervision elements a load shedding scheme can be realized. The working principle is such that unessential load groups are switched off and so the generator can maintain the supply of essential consumers. But the change of frequency reaches trip levels only after a relatively long time caused by the inertia of the genset. For defining the mains frequency in digital frequency meters, the cycle duration is normally measured by means of voltage zero passages because they are relatively easy approachable for measuring purposes. There is actually no difference whether timing convers a complete oscillation or just half. The frequency of the signal analysed can be calculated from this information.

3.1.5 Rate of change of frequency (df / dt ) When measuring the rate of change of frequency, it is not the frequency existing at that moment what is measured, but the speed the frequency changes (df/dt). By this the tendency can be detected before the frequency has changed to such a degree that it is beyond an absolute limit. Additional to the actual measured frequency value, a df/dt relay has to be used so that the previous

Mains decoupling

f
~200 ms

dt

t
df

Vector Surg e

due to reactance XG. This angle is called rotor displacement angle. Prior to decoupling consumer ZL is partly fed by the mains and partly by the own generator. In the following example it is assumed that added to load ZL power is not only being imported from the generator but from the mains as well. If the mains supply is interrupted (t>0), the additional power suddenly imported from the generator is equivalent to the quantity imported from the mains previously. Because of its inertia, the rotor will not abruptly change the speed. Therefore at first the internal voltage does not change. The rotor displacement angle is changed by current change IG IL and so is the phase angle of the terminal voltage which is now not defined by the mains any longer but by the load angle of the isolated system. The change of angle happens very quickly and other effects of this load change (speed drop) reach relative to this a measurable quantity only at a later time. As it can be seen from Fig. 3-4, the generator terminal voltage U1 jumps to a different instantaneous value.
Voltage vector surge u (t) u1(t) u1'(t) Trip

Constant Speed Drop Start O f Sp eed Regulator O p eration

Control Action

Fig. 3-1: Frequency behaviour after mains shut down

Two measuring methods we have realized in our relays will be explained more detailed in chapter 4. These are: df/dt mean value measuring (two-point method) df/dt quasi-instantaneous-value measuring

3.1.6 Vector surge This principle can be explained on the basis of a normal type synchronous generator consisting of internal voltage UP and generator main reactance XG; ZL stands for connected consumers. By C.B. S the mains failure is simulated at t = 0.
t=0 IL

IG

XG

~
M

UP

U1

ZL

G rid

^ t =

Fig. 3-2: Simulation circuit diagram

Fig. 3-4: Oscillogram of the generator terminal voltage at the instant the mains is switched off

The rotor displacement angle is the angle between internal voltage and generator terminal voltage U1
Re Re

t<0

t>0

UP U1

IG
-Im

UP U1
IL
=

U1 / Z L
-Im

Through this the next and also the following voltage zero passages occur at a different time than expected. Consequently measuring the time between the voltage zero passages is the most reliable way for protection devices to detect an angle change. To define the moment this will happen, the relay needs an internal reference. Slower processes, such as changes of frequency and voltage must not cause unintended trippings. Hence it is essential that the reference is always adjusted to the actual mains frequency. A measured time difference t is proportional to a angle change occured. The relay trips if the set value is exceeded. The vector surge supervision is a principle very suitable for quick mains decoupling. By using this method tripping times of about 20 ms (without C.B. operating time) are possible.

Fig. 3-3: Phasor diagrams

In case of power output at the generator terminals, an angle difference between terminal voltage U1 and internal voltage UP develops (Fig. 3-3: t<0).

10

Mains decoupling

3.1.7 Load change (P) Load change supervision is a method very often used in large power stations (P<100 MW) because at a sudden electrical deloading of the generator, the driving turbine has to be protected against overspeed. The protection relay has a direct effect on the turbine regulation and reduces the steam power in case of tripping. For the load change method, however, the measuring connection is more complex because for load measuring three different voltages and three different currents are needed.

are connected in series (interlinked to AND). During healthy operation the current at the mains coupling point is permitted to flow in both directions. The directional current relay is set to the effect to trip in direction public grid if short circuit type currents arise. The undervoltage relay trips if the mains voltage drops under a set threshold. If short circuits occur in the public grid, both relays respond at the same time and the coupling switch is opened. If short circuits occur in the own system, the current relay does not trip. The necessary disconnections in the own system are selectively carried out by other protection elements.

3.2 Combined criteria

In order to clearly distinguish between the different kind of failures in a network, it might be useful to logically combine several of the before mentioned criteria. In this chapter possible combinations are introduced.

3.2.1 Overcurrent and undervoltage (I>& U<) A combination consisting of overcurrent and undervoltage always signals a short circuit or excessive load at the output supervised. In many power export applications an overcurrent relay with inverse current-time characterstic is being used. The characteristic the current relay is based on is defined by the undervoltage relay. If, for instance, the mains voltage drops below 60 % UN at a coincidental I> pick up, a close short circuit is assumed and the relay operated with a faster trip characteristic. In cases where the voltage does not drop below the set value, the short circuit is regarded as having occured further away and a slower characteristic is used for overcurrent trip. Selective shutdown can then be realized by a relay closer to the actual fault. As it is the case with all current-time characteristis, this method is very suitable to disconnect many parallel operating generators selectively from the mains, preventing the entire network to be without voltage at once. This method cannot be used for quick mains decoupling at automatic reclosing because it is too slow due to the trip delay of the current protection. The delay time of the overcurrent detection should not be set too short because there would be a risk of tripping during switching operations and transients in the mains. 3.2.2 Directional overcurrent (I> & U<) By a pure reverse power supervision (see 3.1.3) the operating modes possible are restricted (e.g. only supply into the public grid) because the current is only allowed to flow in one direction. If this restriction is not acceptable for a certain application, mains decoupling can be realized by the combination current direction and undervoltage. To achieve this, the tripping contacts of both relays

Mains decoupling

11

This method makes prompt decoupling of the own system possible. Energy is permitted to flow in both directions. This method is not suitable for decoupling during mains failures with an insignificant drop of the voltage.

3.2.3 Underfrequency and undervoltage (f< & U<) In this combination too the undervoltage is used for detecting a short circuit. The underfrequency element trips if due to mains disconnection the remaining generators are overloaded to a degree that their speed drops. This combination method as well can be used as reliable indication for decoupling during short circuits. Here it should be taken into account that it might take quite a while until the frequency drops noticeably. (See chapter 3.1.4)

12

Mains decoupling

Function of Relays Available

f fN

d t start

Reset Blocking

dt start fe - f T dt TRIP

4.1 Rate of change of frequency

With relay MRF2 two separate measuring methods of rate of change of frequency supervision are possible : Indirect measuring according to the mean value detection of the two-point method Direct measuring according to the quasiinstantaneous value method

df = fe fT f>f T

dt

dt

4.1.1 Mean value detection In this operation mode the relay is only activated after the frequency is above or below an adjustable threshold value fe. If this is the case, a time counter is started (setting value dt in cycles). Should the measured system frequency be also above or below the second threshold value, i.e. tripping value fT, within time t, the MRF2 trips immediately. Tripping value fT results from setting df, fe and dt : df = fe fT fT = fe - dft t

Fig. 4-2: Operating principle of df/dt measuring

4.1.2 Quasi-instantaneous value From every new measuring value of the cycle duration, the MRF2 detects the frequency gradient df/dt of the mains voltage. Every 10 ms (at fn = 50 Hz) this method presents a new df/dt value and that measuring value is called "quasiinstantaneous value" for this instant of time. To evaluate the rate of change of frequency the MRF2 consideres more than two frequency values. Multiplex evaluation inevitably entails delay of the fastes tripping time (the more exact, the longer it takes). Since, however, duration of the multiplex evalutation can be parameterized, accuracy and speed can be adjusted optimally to the prevailing mains conditions. This special measuring method provides a high degree of tripping reliability and consequently high stability during switching actions. A short-term oscillation of the mains frequency, as can be caused by switching actions, does not result in trippings. Depending on the setting, the total switching off time at mains failures is about 60 ms - 80 ms (without C.B. operating time). This measuring method is very resistant to switching transients and associated short-term frequency oscillations.

If tripping value fT is not reached withing t, then there is no trip. Only if the frequency rises above the first threshold value fe, the MRF2 is automatically set back. Fig. 4-2 shows the frequency behaviour with tripping and without tripping.
f < fe
t =dt

f
f < fT

&

&
1

T R IP

Fig. 4-1: Block diagram of the mean value method

Mains decoupling

13

4.2 Vector surge measurement

The MRN2-1 measures the cycle duration and at each voltage zero passage new measuring is started. In order to predict the time of the next voltage zero passage, the relay reproduces an internal frequency replica of the mains voltage. This reference replica is constantly adjusted to the actual frequency. In case of a vector surge (fig. 3-4), the reference replica and the mains frequency are not identical any more. After a vector surge, the next zero passage is either earlier or later. From the difference between reference time and the measured time the angular shift of the mains voltage can be defined. In theory this method would enable the relay to detect a vector surge after about 10 ms, but the MRN deliberately was not designed to realize the quickest possible tripping time. The reasons for this are: As demonstrated by practical measurings, a vector surge process (e.g. 10) can spread over several measuring periods (e.g. 4 and 6) through several small consecutive vector surges. A "one-period relay" could not detect this trailing vector surge if it is set to 8, for instance. Because of load connections and synchronous switching actions, shift of a single zero passage may occur . This shift can be identified by a sequence of two contrary vector surges. A normal relay would trip in such a case because it does not consider the sign. If, for example, the contacts of a circuit breaker react time-delayed, the measured vector surges are not significant and have to be considered as more accidental. In this case as well the multievaluation detects the process clearly. Only after three successive angular measurings, the MRN2-1 gives a tripping impulse. By this method a high reliability of the trip decision can be granted and the relay is still fast enough to be used for quick mains decoupling. The trip command is given by MRN2-1 about 60 +/- 10 ms after a vector surge has been detected. The tolerance of abt. +/- 10 % can be explained by the uncertainty of the vector surge within the half-cycle.

Undervoltage lock out The vector surge function in the MRN2-1 can be blocked in case of undervoltage. The relay is neither triggered by failure of one phase or several phases of the measuring voltage, nor by a short circuit in the mains during which the voltage significantly drops. The undervoltage threshold value of this function can be set. This function has been provided because normally a vector surge relay is not able to detect the fault location: the voltage in the mains is normally very much the same. In cases where short circuit protection is realized by different selective protection systems (e.g. distance protection, line differential protection etc.), the vector surge relay can be blocked. If not necessary, the undervoltage lock out can be switched off. In this case the relay trips at any recognizeable angular difference. Trip logic for vector surge measuring The vector surge function of the MRN2-1 supervises vector surges in all three line to line voltages at the same time by phase conductor voltages. Tripping of the relay can optionally be adjusted for single-phase or three-phase vector surge. The single-phase variant should be chosen if single-phase failures, e.g. conductor fault etc., shall result in tripping. Whereas the three-phase variant is of advantage for applications where only three-phase failures need to be detected (complete loss of mains in all three phases). Blockade after measuring voltage connection Right after the measuring voltage has been applied, the MRN2-1 cannot identify correct phasor position. Apart from this, transient and autooscillation processes may influence the measuring voltage at this instant for a certain time. To prevent that the respective mains section is disconnected instantly by false tripping due to angular difference, the MRN2-1 blocks the vector surge for 5s after measuring voltage has been applied.

14

Mains decoupling

Time blocking by the external blocking input Generator synchronization can be used as an application example for this blocking function. The vector surge function is only adviseable for mains parallel operation, because in single operation any load change is followed by a vector surge; therefore the vector surge function must only be activated after the generator has been synchronized. This can be achieved if an aux. contact (e.g. of the C.B.) which applies voltage to the blocking input of the MRN2-1, when the C.B. is open. When the generator is synchronized, the blocking signal is switched off and the vector surge function available after 5s. Min./-max.-polling By this function always the largest and smallest angular difference is stored; the measured values are not subject to the multi-evaluation, hence after tripping, for instance, the value in the max. storage can be different to that in the trip storage. For assessment of the mains quality, however, this value is very informative because by means of calculation it reflects unfiltered the "noise" of the mains phasor. Even if there was no tripping, the stored measuring values indicate the identified angular differences in the mains. During commissioning work these measuring values can be used to ascertain suitable setting limits. Border cases: Although nearly under any operational conditions in mains/generator parallel operation vector surge relays grant a reliable and quick detection of mains failures, the following border cases should be considered: a) No or only insignificant changes of the load flow at the mains coupling point when a mains failure occurs. Such constallation could be found in peak load plants or CHP stations where the load flow between the power stations and the public grid can reach very small values. In order to identify a vector surge at the mains parallel operating generators, an active load of at least 15 - 20 % of the rated load is necessary. These are experimental values. If the active load at the mains coupling point is regulated to minimal values and then a "high-resistive" mains failure occurs, i.e. a failure with insignificant load changes, neither will there be a vector surge nor changes in load and frequency. Consequently this failure is not recognized. This case only occurs if the public grid is disconnected near a power station and so the generators cannot additionally be loaded with the remaining grid. At more distant mains failures the synchronous generators are loaded abruptly due to the larger content of the residual grid and a vector

surge occurs. Here mains failure detection is certain. If such a case occurs, the follwing should be observed: When a mains failure is not detected, i.e. the intertie C.B. remains closed, the vector surge relay reacts at the first load change causing a large enough vector surge, and then opens the mains C.B.. If, however, in the meantime the frequency tolerance of an overfrequency or underfrequency supervision unit is exceeded, the vector surge relay opens the mains C.B. earlier. For detection of high-resistive mains decoupling it is also possible to use a "zero current relay ", i.e. a minimal current relay (setting abt. <3% IN). This relay has an adjustable time delay which is necessary for regulating processes when the current at the mains coupling point reaches zero for a short time. At high-resistive mains failures the inter-tie C.B. is tripped by the zero current relay after elapse of the time delay. In order to prevent asynchronous connection, automatic reclosing of the public grid should be blocked at least for this time delay. Furthermore it is also possible to regulate the load at the mains decoupling point such that always an active load flow of 15 - 20 % of the generator load is granted. b) Short circuit type loading of the generators at remote mains failures. At each remote mains failure, the remaining public grid causes abrupt short circuit type loading of the generators in the power station. The vector surge relay recognizes the mains failure within abt. 60 ms and switches off the inter-tie C.B. So the complete time for switching off is abt. 100 150 ms. If the individual generators have an extremely fast acting short circuit protection, e.g. for detection of di/dt, the generator C.B. might unselectively switch off the generators. This is undesirable because then the power supply for the own consumption is at risk and in some cases synchronized changeover after AMF operation is only possible after the overcurrent protection has been manually reset. To avoid such a situation, the generator C.B.s have to be provided with delayed short circuit protection. The time delay should be such that at least mains decoupling by the vector surge relay can be effected.

Mains decoupling

15

Example for Relay Applications

5.2 Combined heat and power plants

5.1 Embedded generator with possibily of island operation

GRID
CHP
P ro t. de v ic e

G rid Cu toff

CHP
P ro t. de v ic e

CHP
P ro t. de v ic e

oth er Lo ad

Public P rivate

Pr o t. D e v ic e

In ter-tie C.B .

Fig. 5-2: Group of CHPs with a major share in the joint power supply for the region.

Int. Lo ad

G enerator Switch

This example shows a number of individual CHPs spread over a region. Each of these power stations has its share in the joint supply of the region. Such configurations were mentioned in chapter 3.1.2. Important is that in case of failure not all units which can measure the failure may stop supplying immediately. If there are too many units decoupling from the intrasystem, the stability of the remaining system would be at risk. Loss of mains relay MRN2-2 Frequency gradient The parallel generators cause a small relevant reactance. Here a df/dt relay is recommended. Protection parameter df/dt = 2 Hz/s dt = 4

Fig. 5-1: Power generating system in parallel operation with single operation features

This example probably shows the most frequently used kind of power generating system, consisting of a generator and own consumers (e.g. an industrial works). The power generating system is connected with the public grid by means of inter-tie C.B. Conditional on the capacity of the generator, by this example the entire spectrum of failures and decoupling criteria (chapter 2.3) can be demonstrated. Generator data SN=14.7 MVA UN=10.5 kV Loss of mains relay MRN2-1 Vector surge In case of mains failure a large change of generator load is expected. The vector surge is large also because of the high relevant reactance. Protection parameters U< = 90% UN, tU< 1.2 s U<< = 85% UN, tU<< 0.1 s U> = 110% UN, tU> 1 s U>> = 115% UN, tU>> 0.2 s f1=48.5 Hz, tf1= 500 ms f2=49.0 Hz, tf2= 1 s f3=51.0 Hz, tf3= 1 s = 8

16

Mains decoupling

5.3 Long closed-loop network

Long O pen Air Transm ission Line

5.4 Wind power station

As y nc hro no us ge ne ra to r

G
Long O pen Air Transm ission Line Long O pen Air Transm ission Line

G
Low Inte rna l C o ns um ptio n

Fig. 5-4: Wind power station with asynchronous generator

U1

U2

Long O pen Air Transm ission Line

In wind power stations very often asynchronous generators are used. For fast mains decoupling, this generator type, however, can be treated like a synchronous generator.

Mains Feed

Fig. 5-3: Meshed overhead line

Such a closed-loop arrangement can be used in widely branched networks. The consumers are spread over the loop which may have a scope of several 100 kilometres. When for operational purposes the disconnecting point is opened, U1 and U2 could turn by some angular degrees because of the line impedance. In this case a vector surge relay connected to U2 may trip because of these switching actions. Here a rate of change of frequency relay for loss of mains protection would be more suitable.

Mains decoupling

17

5.5 Embedded generators as UPS

(UPS = Uninterrupted Power Supply)
Thy ris to r S witc h P ublic G rid Impo rta nt Load

G
G G G

Fig. 5-5: UPS system with extremely fast mains decoupling features

If the security of power supply for the own consumers has priority, the decoupling time has to be optimized everywhere. This specificly applies to power generating systems which are used, for example, as UPS system in a computer centre. Our example shows an arrangement where several small diesel gensets can redundantly secure the entire supply. For mains disconnection a vector surge relay without multi-evaluation and a thyristor switch are used. Normally the gensets are balanced that virtually no current flows via the coupling switch. If the thyristor switch does not receive any ignition pulses then there is no current in any phase at the next zero passage, i.e. after 10 ms the latest. By using the single evaluation vector surge method, the theoretical tripping time of 10 ms can nearly be obtained. As soon as the protection relay recognizes a vector surge, further igniting of the thyristors is blocked. Mains disconnection can then be realized in abt. 10 - 20 ms. Although this variant of mains decoupling is not completely trip safe, but in the interest of continuous power supply (computer centers / process control) fast decoupling is permitted, even if there is the risk that unintentional trippings may occur. Provided the generator has enough capacity, decoupling after such a short time is sufficient to ensure uninterrupted operation of the computer center.

18

Mains decoupling

Summary
As to planning of protection schemes for mains decoupling the following guiding principles are outlined: The amplitude of the vector surge is defined by the active load change at the generator during mains failure. For strong active load changes, the vector surge relay is very suitable to ensure fast and reliable decoupling. A large relevant reactance between the generator and the point the vector surge relay is connected increases the measurable vector surge. This should be taken into account when planning the relay installation place. In block operation, the protection relay should be connected at the mains side of the unit-transformer. The higher the share of motors on the load of the isolated system, the higher the actual generator capacity available at the instant of mains failure. Simultaneously to the increasing share of motors on the load of the isolated system, the relevant impedance decreases. For such an application a rate of change of frequency relay can be recommended. Very often a combination of vector surge and rate of change of frequency could be adviseable because by this the range for detecting different mains failures conditions is extended and so protection of generators becomes more enhanced.

For captive power generating systems operating parallel with the mains it is very important that in case of mains failures they are quickly decoupled from the public grid. Different fundamental viewpoints of the different interest groups and operational modes have to be taken into account when planning the protection scheme. The time range of interest is 200 ms after a failure has occured to be sure that the decoupling process is still fast enough in the event of automatic reclosings. Some of the conventional protection criteria (measuring of voltage and frequency) are not suitable here because mains failures are not quickly enough detected by those. Nearly in all cases the better solution is to use the methods of rate of change of frequency supervision (df/dt) and vector surge (). By these most of the failures which may occur are covered. Appropriate protection relays are available and these were specifically designed for fast mains decoupling. When knowing the customer related system data, calculation for relay adustment is possible without problem. This essay does not treat the theory required for the mathematical estimation of the processes in the event of a mains failure. Being aware of this background it is possible to determine the curve of the voltage, of the phasor position and the rate of change of frequency df/dt at a mains failure on the basis of only few characteristical system data. For all practical purposes, these calculations result in sufficient accuracies verified in theory and in practice. They simplify the selection of the proper protection relay as well as the determination of setting parameters.

Mains decoupling

19