Equipment and procedure to determine fault location and fault resistance during phase to ground faults on a live network

The subject of the invention is an equipment and procedure to determine fault location and fault resistance during phase to ground faults on a live network with a resonant or high impedance neutral grounding. The solution subject to the invention can be advantageously used on live networks consisting of overhead lines, cables or both, with isolated neutral or neutral grounding via impedance or resistance, in order to quickly and accurately determine fault location and fault resistance in case of either stable or intermittent ground faults.

It is a well-known fact that several solutions have been elaborated for determining fault locations and fault resistance. The most widespread alternatives are so-called fundamental frequency (50 Hz or 60 Hz) methods, however, all such methods are deficient in that at least two different operating states are required, which cannot be concurrently established and, consequently, that changes in certain quantities at fundamental frequency between the two measurements impact on the measured quantities thus making computation overly inaccurate for example: in the case of power lines running on the same set of poles, the load change occurring on the non-faulted line induces a voltage in the faulted one thus falsifying the measurement being done thereon, during a short-circuit existing on the medium voltage network (while measurements are being done), any short-circuit happening on the high-voltage side results in a zero sequence induction in medium voltage thus falsifying the measurement being done on the medium voltage side, the load varies on the faulted or non-faulted branches.

An additional deficiency of calculation involving fundamental frequency quantities is that, for the benefit of accuracy, measurement needs to be performed for a long time (several seconds) due to the creation of various operating states so, during measurement, there is a considerable probability of the above-mentioned phenomena occurring,

the fault location's impedance value changes, which likewise usually degrades accuracy in fault location determination.

Among the documents a deterministic direct method described in patent description number US 2007/0124093 represents for example the state of the art, however, this can only be used in the case of dedicated cable lines. It relies on computer simulation and assumes knowledge of all parameters as well as homogeneity over the entire measured length. This known solution is deficient in that it considers only fundamental frequency

(e.g. 60 Hz) signals, furthermore, it does not take into account any measurement uncertainties, under realistic circumstances probably works at a high error rate due to computer simulation being used, and is suitable only for determining cable faults so it does not facilitate use on overhead lines or mixed networks.

A procedure described in patent description number EP 1304580 likewise represents the state of the art and allows calculation of the fault location distance in the case of single- phase to ground faults. The procedure employs a fundamental frequency signal. Measurement is based on forced changing of the zero sequence impedance, utilizing measurement results obtained for the two different values thereof. Consequently, the fault has to exist for a relatively long duration and, due to the use of a lumped parameter model, experience shows that its accuracy is lower than if distributed parameter lines were used. Additionally, the procedure does not take into account positive and negative sequence capacities whose ignorance likewise reduces accuracy.

The procedure described in patent description number DE 10143595 likewise represents the state of the art, which involves detection of single-phase to ground faults on radial distribution lines. The solution has the same disadvantage as the method described in patent description number EP 1304580.

Patent description number EP 1089081 likewise deals with fault location which essence is that a loop is created for a short duration using the defective and a non-faulted branch whenever there is a fault and that the fault location is determined based on information

(numerical measurement data) that can be obtained from the loop configuration. The

deficiency of this known solution is that conditions for creating a loop are not given in every case, as mentioned in patent description number DE 10143595, furthermore, due to the usage of fundamental frequency (e.g. 50 Hz) quantities, the disadvantages already summarized in the introduction also appear in this solution.

A procedure described in patent description number WO 97/08562 represents the state of the art, which is used to determine the location of single-phase to ground faults in electric distribution networks by processing transient signals. This solution is deficient in that the transient process depends on the fault location's resistance, which renders this procedure inaccurate, therefore, its use is not expedient, which is even referred to in several later patent descriptions.

The procedure and equipment described in patent description number JP 4140016 represent the state of the art, which is used to locate ground faults and to determine the distance thereof. This method involves estimation of the fault location resistance based on fundamental frequency zero sequence voltage and current and fundamental frequency phase voltage, therefore, the result is expected to be inaccurate. In addition, the disadvantages described in the introduction also appear in this method.

Fault location on electric lines described in patent description number DE 3016223 likewise represents the state of the art and is based on the travelling wave reflection method. The deficiency of this known solution is that measurement needs to be first conducted on a non-faulted, switched off and non-operating line, for example during commissioning, which makes subsequent installation circumstantial and the measurements need to be repeated for example whenever a new side-line is installed, furthermore, a generator suitable for creating and coupling impulses shorter than 2 ns is required, which likewise makes practical use difficult and significantly raises costs.

The invention envisaged to eliminate the deficiencies of known solutions and to create an equipment and procedure involving its use to determine fault locations on live networks consisting of overhead lines, cables or both, with isolated neutral or neutral grounding via impedance or resistance, preferably fault locations in medium voltage

overhead line networks and the resistance of the fault location, furthermore, for the benefit of continuous power supply, these be implemented economically, employing a simple design, adapted to the substation protection automation system, more accurately than with the known solutions, employing real-time measurements of relatively short duration and independent from the disturbing effects of external circumstances to a great degree.

The solution subject to the invention is based on the recognition that, if the parameters of an electric network are available and stored in a data processing unit while constantly monitoring the rms value of zero sequence voltage and if a preset value is exceeded, the injecting unit is controlled by a measuring and controller unit so that at least one, preferably two, measurement currents concurrently lasting for a short time, preferably 0.5 to 1 seconds are injected into the star point at predetermined frequencies different from the fundamental frequency and its odd-order harmonics, and concurrently with injection the measuring and controller unit is used to selectively measure the zero sequence currents of the branches and the zero sequence voltage, all at measurement frequencies or at injected signal frequencies then the measured data are transferred to a data processing unit in which the data are averaged as needed and then parameter matching is done, during which the two unknown parameters of the model (preferably a distributed parameter line model composed in consideration of the network's electric parameters), namely the distance of the fault location and value of fault resistance, are varied so that the difference between the zero sequence quantities at the measurement frequencies or at the injected signal frequencies simulated using the model and measured by the measuring and controller unit is within the predetermined tolerance, then the obtained pair of values will provide the distance of the fault location and the fault resistance so the solution subject to the invention achieves its objectives.

In its preferred embodiment, the equipment subject to the invention is implemented as described in Claim 1. The individual examples of embodiment can be implemented according to Claims 2 - 4. In its preferred embodiment, the procedure subject to the invention is implemented as described in Claim 5. The individual procedure variants are described under Claims 6 - 9.

The solution described in the invention is demonstrated in detail through figures where: Figure 1 shows a schematic drawing of the equipment subject to the invention along with the substation, Figure 2 shows a basic circuit diagram of the injecting unit, Figure 3 shows the coupling/filtering unit's circuit diagram.

Figure 1 shows a schematic drawing of the equipment subject to the invention along with the substation. Among other things, the measurement system is connected to zero sequence current transformers Ia, the I _0, i _, ... I _0,n zero sequence currents of the secondary windings thereof, the zero sequence voltage-signal conditioner Ib, the U ₀ zero sequence voltage generated on the output thereof, the star point impedance Z between the star point and ground, and the substation 1 possessing a dispatcher computer Ic. The input 21 of the measuring and controller unit 2, on the one hand, connects to the secondary windings of the substation's 1 current transformers Ia, and on the other hand, to the output on the zero sequence voltage-signal conditioner Ib, its data output 22 connects to the data input 31 of the data processing unit 3, the controller's output 23 connects to the controller input 41 of the injecting unit 4. The injecting unit's 4 output 42, on the other hand, connects via a coupling/filtering unit 5 to the star point. The data processing unit 3 has a result output 32 which preferably connects to the substation's 1 dispatcher computer Ic. (The connection is marked by a dashed line in the figure.)

Figure 2 shows the circuit diagram of the injecting unit 4. The injecting unit 4 preferably has a network unit 4a, an intermediate DC section 4b connected to it, an inverter 4c consisting of switching units connected to the DC section 4b, while the inverter 4c possesses two high-frequency outputs 4c i and 4c ₂ . (The controller input 41 of the injecting unit 4 not marked in the drawing controls the switching elements of the inverter 4c.) The network unit 4a is preferably a three phase diode bridge with a DC section 4b implemented using a capacitor C, while the inverter 4c is built from four semiconductor switching elements.

Figure 3 shows a circuit diagram of the coupling/filtering unit 5. The coupling/filtering unit 5 preferably has a high frequency filter 5a, a DC isolating capacitor 5b connected in series with it, an isolating and coupling transformer 5c connected to it, a fundamental frequency decoupling filter 5d, and connected to it at least one, preferably two series filter circuits 5e tuned to the injected signal frequencies. The drawing shows both high frequency outputs 4c i and 4c ₂ of the injecting units 4 connected to the inputs of the high frequency filter 5a. The number of series filter circuits 5e depends on the number of injected signal frequencies used, and it is clear from the figure that this number is two in a preferred embodiment. The filter circuits are composed of an inductive element and a capacitive element. The coupling/filtering unit 5, on the one hand, is responsible for assuring minimum power loss in connection with the injected signal, in other words that the largest possible portion of the injected signal reaches the network by using an injecting unit 4 with a rated power as small as possible and, on the other hand, to prevent the high carrier frequency signal of the injecting unit 4 from reaching the network, and furthermore to minimize inrush currents and permanent loading of the injecting unit 4.

The operation of the equipment subject to the invention in view of the above-mentioned figures is described as follows: First of all, the equipment is installed in the substation 1 to be examined which has been powered off.

The zero sequence voltage-signal conditioner Ib is preferably designed so that it generates the zero sequence voltage U ₀ from the voltages at the secondary windings of the voltage transformers connected to the bus bar.

In order to improve measurement accuracy, the phase shift and magnitude error of the voltage- and current transducers are expediently measured then the network can be powered up again.

The values (which are in cases current-dependent) required for correcting phase shift and magnitude error are stored in the data processing unit 3.

Afterwards, the network's typical electric parameters (number of branches, location of side-lines, line section lengths, the specific positive and zero sequence line inductances at the measurement frequencies, the specific positive and zero sequence line capacitances, the specific positive and zero sequence line resistances at the measurement frequencies, the specific earth resistance at the measurement frequencies, the rated transformer power, the transformer drop and the short-circuit power at the substation if known) are stored in the data processing unit 3.

Having performed the above mentioned preparations, the equipment is ready for operation.

After being switched on, the equipment would constantly monitor the rms value of the zero sequence voltage U ₀ . If this value exceeds a preset value, the injecting unit 4 is controlled by the measuring and controller unit 2, so that it injects a measurement current for a short time, preferably lasting 0.5 to 1 seconds, into the medium voltage transformers star point on at least one, preferably two predetermined frequencies.

Injection can be performed at just one frequency other than the fundamental, however, the fundamental components are also to be measured and processed in this case. By using this method, the "injected signal frequency" hereinafter will be understood as a frequency other than the fundamental.

The deficiency of this method is that the procedure is slow and is sensitive to fundamental network noise, therefore, the fault location distance / and the fault location resistance R thus determined will be inaccurate.

The simultaneous use of at least two injected signals on a frequency other than the fundamental and its odd order harmonics (for example 400 Hz and 500 Hz) is therefore expedient. By using this method, the "injected signal frequencies" hereinafter will be understood as the frequencies other than the fundamental.

The fundamental frequency and the injected signal frequency (or frequencies) shall be understood as 'measurement frequencies.'

Concurrently with injection, the measuring and controller unit 2 is used to selectively measure branches' zero sequence currents I ₀ , i ... I _Otn and the zero sequence voltage U ₀ at the measurement frequencies (fundamental and at least one other frequency) or at the injected signal frequencies (at least two frequencies other than the fundamental).

The measured data is transferred to the data processing unit 3 where averaging is performed whenever needed.

The previous two steps (measurement and data transmission) can be speeded up significantly if the zero sequence current in only the faulted (ground fault) branch is measured (at the measurement frequencies or at the injected signal frequencies) and transferred to the data processing unit 3. In this case the defective branch is determined by first using the measuring and controller unit 2 to measure only the fundamental components of each branch's zero sequence currents I _Oj i ... I _0^n and only the fundamental components of the zero sequence voltage U ₀ and transfer it to the data processing unit 3. Then the zero sequence active power per branch calculated from this data are compared and - in the case of a measurement assuming positive active consumed power - the one with the greatest absolute value and negative zero sequence active power will be the faulted branch.

A parameter matching procedure is done after averaging the data during which two unknown parameters of the modeled substation network composed in view of the network's electric parameters, namely, the fault location distance / and the fault resistance R, are varied so that the difference between the zero sequence quantities simulated using the model and measured by the measuring and controller unit 2 (at the measurement frequencies or injected signal frequencies) is within the predetermined error threshold.

The error criterion for parameter matching can be expediently defined as follows:

N

H = Y (*)

^J Q, measured - Z, 0,model , where k=\

H is the error

N is the number of measurement frequencies used k is the serial number of measurement frequencies (a number between 1 and λO

^o,mea _s ur _ed is the zero sequence driving point impedance of the faulted branch on the k ^th measurement frequency calculated based on measurements,

^o, _model is the zero sequence driving point impedance of the faulted branch on the k ^ih measurement frequency calculated based on simulation,

and

rjik) χ(k)

0, measured , where

U o, measured is the k measurement frequency component of the measured zero sequence voltage τ(k) . _th

J- o, measur _ed is the k measurement frequency component of the faulted branch's measured zero sequence current τ r(k)

^U 0,model is the k ^l measurement frequency component of the faulted branch's simulated zero sequence voltage

¹ 0,model is the k ^x measurement frequency components of the faulted branch's simulated zero sequence current

The results of parameter matching will be the value of the ground fault location distance / and the fault resistance R.

The two above-mentioned results are made known to the operators of the 1 substation via the 32 result output of the 3 data processing unit (for example on a display, printer or communications port).

The equipment subject to the invention was tested in a series of site tests done on a preferred implementation in a 120/20 kV substation, in a preinstalled fault location with given fault resistances R.

An 18.95 kilometer overhead line and a 649 m ground cable are located between the bus bar and the fault location. Two sets of measurements were performed using the same fault location distance / with a fault resistance R of 100 ohms and 6,000 ohms.

Once the ground fault occurred, the measuring and controller unit 2 was used to modulate the injecting unit 4 at 400 Hz and 500 Hz. The duration of the measurement was 600 ms.

Measurement results:

It can be declared according to the measurement results that the fault location distance / can be determined with an accuracy of no more than 1% while the fault location resistance R can be determined with an accuracy of more than 10%.

The solution subject to the invention has implemented its objectives and affords the following advantages:

the fault resistance and ground fault location of a live network consisting of overhead lines, cables or both, grounded via impedance or resistance or having an isolated neutral, can be determined in real-time, based on a very short measurement (less than one second in duration), - a fault location's distance and the fault resistance value can be determined rather accurately with an accuracy within 1 or 2% regarding the fault location's distance and 10 % regarging the fault resistance, thanks to the fast computation , it is also suitable for locating intermittent short- circuits (such as caused by trees), - due to the use of frequencies other than the fundamental, the solution is insensitive to changes or transients in network operation, due to the use of frequencies greater than the fundamental, measurement durations (typically 0.5 to 1 seconds) less than those of known alternatives are sufficient, - it can be adapted well into existing substation automation systems, it consists of a simple construction, it can be manufactured economically.