AT500004B1 - METHOD FOR DETECTING AND LOCATING AN EARTH LOCK - Google Patents

Description

2 AT 500 004 B1

The present invention relates to a method for detecting and locating a low-resistance or high-resistance ground fault in electrical power supply networks.

In the publication WO 96/27138 A1 a method is presented with a working at the mains frequency power supply. To determine the conductor-earth capacitance, two measuring cycles are required during which the network or the contact resistance at the fault location must not change. In practice, it has been shown that especially in the case of high-impedance faults, the contact resistance changes greatly due to charring phenomena at the fault location. The required measurement cycle for obtaining the reference values can take up to several minutes, because, among other things, settling times must be waited. Furthermore, in this method, with respect to the essential quantity E1, which corresponds to the voltage of the positive sequence system, assumptions are made in terms of size and direction, which often do not apply.

The method is not suitable for low-resistance ground faults, since the mains-frequency power supply does not achieve a sufficiently large voltage change. This makes it impossible to determine the phase-earth admittance in the case of low-impedance errors.

If the displacement voltage is the same as the voltage of the power supply, the connection causes the " current injection " no change of the displacement voltage, so that even in this case, a calculation of Ya and Yu can not be done.

With the method compensation or suppression by crosstalk of the positive sequence in the zero current is not possible. Furthermore, it is not possible to compensate for the phase splitting, which usually occurs in the frequently used meshed networks.

In the method of WO 92/18872 A1, the admittances of each individual outlet before and after the change of the displacement voltage are determined from two pairs of measured values. A nonzero difference indicates the erroneous exit. Since the voltage change is purely passive by changing the tuning of the network, it is necessary that a sufficiently large displacement voltage exists. The method does not work for symmetric networks, because then a determination of the zero admittance is not possible. Due to the necessary adjustment of the Petersen coil in case of a ground fault, a lot of time is needed for the adjustment of the Petersen coil. During this time it is assumed that the contact resistance at the point of failure does not change. In the case of a very sensitive setting, the Petersen coil is started to scan every time UQ changes slightly.

In the method according to WO 98/20356 A1, only the displacement voltage U0 and the clamping voltage ϋλ are measured in the substation. The high-impedance ground fault is detected by comparing the direction of Z0 with the positive voltage. A segregation-selective detection of the ground fault is not possible. In addition, the zero impedance Z0 is required, which, however, only provided via network models to the relay. An independent detection of a change in the switching state is not possible. It can be shown that in case of imbalance of the network, with this method a changed line length is interpreted as a change of the asymmetry and thus as a high-impedance ground fault. A check of the phase-earth admittance is absolutely necessary before the display of a high-impedance ground fault in order to prevent an overreaction of the protective relay.

In the published patent application WO 99/10753 A1, a method for detecting high-resistance ground faults is presented. For the calculation, the unknown and according to the published patent not easily measurable zero impedance Z0 is required here. According to the published patent, new values for Z0 are only available after a ground fault. Switches in the healthy network are not detected or an independent detection of a change in the switching state is not possible. It can also be shown here that in the case of asymmetry of the network present, in this method a changed line length is interpreted as a change of the asymmetry and thus as a high-impedance ground fault.

In DE 44 28 118 A1, a named earth fault location in electrical networks with a ground fault coil, a current of specific phase position is generated in the ground fault extinguishing coil by secondary detuning of the earth fault coil setting, which is used for further calculation or evaluation. WO 2002/015355 A2 also shows a method and a device for detecting and locating only high-impedance, single-pole earth faults, wherein the application is disclosed with the supply of an auxiliary signal with the mains frequency and is based on a substantial compensation of the neutral point displacement voltage. Even with these methods, el. Coils, such as are increasingly prevalent in networks, can not be sufficiently considered or recognized.

A check of the phase-earth admittance is absolutely necessary before the display of a high-impedance ground fault in order to prevent an overreaction of the protective relay.

In summary, the known methods for the detection of single-pole high-impedance earth faults fail in the normal operating conditions in the network, because assumptions are made which are not sustainable in practice. Essentially, the crosstalk of the changing load current to the zero system, the variable resistance at the fault location or the high impedance to be determined zero impedance disturbing. In practice, the sensitivity must be drastically reduced, i. in the range of a few 100 ohms, so that no false positives are caused.

All the problems mentioned above are eliminated with the new method, since the determination of the sum admittance by measuring at a frequency other than the mains frequency. The load-dependent crosstalk of the fundamental is thereby filtered out in the determination of the sum admittance. The influence of the faulty measurement by the open delta winding, in which the sum of three large voltages is formed, is greatly reduced, since the two frequencies are fed as zero voltage and measured in phase with earth. An interfering superposition by a much larger voltage does not occur at these frequencies. In addition, the two frequencies are measured and evaluated at the same time. A temporal constancy of the network and the crosstalk for a period of a few seconds, as required by the other methods, is not required in the new method. A measuring period of 200 to 240 ms is usually sufficient. Likewise, the disturbing influence of the phase splitting can be eliminated with this method. As a result, selective detection of high-impedance ground faults is possible for normal operating states of the network.

As high-impedance ground faults, transitional resistances of more than 1 kOhm are already referred to in general usage, because they can no longer be detected with the usual methods. For example, high impedance ground faults occur when a fallen tree touches a line (40K ohms to 100K ohms), a ladder rope falls on dry sand, dry rock, snow, or ice after a rope break, or after a broken wire break, at which a consumer returns Conductor Although the earth touched low impedance, but seen from the supply side of a high-impedance ground fault, because the impedance of the load-side transformer and its load state enters into the ground fault with.

High-impedance ground faults, however, are very dangerous. Although the current at the fault point is reduced by the high resistance, very high stepping and touching voltages result for the same reason.

The above-mentioned methods work with the assumptions made, but they are too idealized in normal network operation and are not tenable. 4 AT 500 004 B1

The new method measures at a frequency different from the mains frequency. The power-frequency components are filtered out and do not affect the determination of the Υχ_x values of an outgoer.

By simultaneously feeding two currents at two different frequencies and simultaneously evaluating the measurements for these two frequencies and the line frequency, the calculation of Y ^ _x and the imbalance Yu_x of the departure is very fast and at the same time. It is not necessary to use consecutive measured values or network states for the calculation.

By a parameterization maximum values for the change of the displacement voltage y0 caused by the current source or maximum current values for the current source can be specified. By monitoring the displacement voltage, the size of the injected currents is adjusted so that on the one hand, a sufficiently large displacement voltage arises even in large moods or on the other hand, the limits for U0 is not exceeded for very small networks.

As a triggering criterion for a calculation of the summation admittance may be a change in the power frequency shift voltage.

For symmetrical networks, a change in the network does not necessarily lead to a change in the residual voltage. In this case, a mixed stream is constantly fed and the monitoring of the cable lengths of a departure is constantly.

In the case of mains-frequency power supplies, very complex synchronization circuits must be built for decentralized relays in order to synchronize the measuring times with the switching state of the power supply. In the new method, the synchronization and the detection of changes in the switching state automatically results.

The infeed can be made via a power development winding of the Petersen coil or via an auxiliary transformer. The voltage U0 can be done directly at the outlet or on the busbar with the known methods of open delta winding. If the voltage U0 at the busbar is measured, an unambiguous assignment of the voltage to the respective outgoer x must take place. The zero currents can be measured with the aid of a cable conversion transformer or with the help of the Holmgreen connection in the outgoing junction x.

According to the invention, the method for detecting and locating a ground fault is characterized by the following steps: A) At time U, a measurement and calculation of the sum admittance to ground and a breakdown of the unbalance to the individual phase-earth admittances occur from the following Steps: a) Feed a composite stream into the null system of the deleted network. b) Where the current from the superimposition of the current with the amplitude In and the frequency / i and the second current with the amplitude / e and the frequency f2 composed. c) measuring the displacement voltages L / 0ji_x and L / 0_t2_x according to known methods for the two frequencies per outlet x or at the busbar with clear assignment to the outlet x. d) measuring the sum currents k_f1 _x and! χ_, 2χ according to known methods according to magnitude and angle for the two frequencies per output x. The angle refers to the associated 5 AT 500 004 B1

Displacement voltage L / 0_fi_x or U0_f2_x of the feeder x. e) Determining the sum admittance per departure x for both frequencies: = -M1 - «- or y = -Σ-'2- *! ~ tf 1 X II U ^ -W- I_f 2 II ~ ~ -0 ft x ¥ -0 f2 x (14)

Yr with Y-Y + Y + γ -τ_ίi_x -2 n x -3_n_x 1 (15) G1x + MC1x + 7 G2x + MC2x + G3x + MG3x +7

Gx +} ωλΟΕχ +7 • Ex

Y -Y + Y + Y -I f2 x II f2 x -2 f2 x -3 f2_x G1x + 7ö% G1x + 7 1

yes> 2L G2x + MG2X +;

Gx + y <y2G £ x + 1x ^ / 1 ^ jo) 2L2x ^ G3x + MG 3x (16)

Ex f) The effective component Gx per output of the equivalent circuit results directly from the real part of the sum admittance. y £, "» ree / jjT ^ -i-UGfc + Gs. + Gs.'ß, (17)

[-O_n_xJ g) The calculation of the earth capacitance Cb, and the inductance of any extant particle coil Z_Ex in the outflow, takes place from each of the two imaginary parts by transformation (18) (19) (20) Υυ_η_χ = · ΐ ^ 9 {Υυ_η_χ} or Y ^ _, 2 x = -imag ^ f2 x) GEx = Y &amp; ! 2 χω2 ~ ^ 1 / ft χω \ 2 2 - <y2 L = _1_EX ωΐ (^ Σι H x + <ö1GEx) h) Calculation of the sum admittance per output for the line frequency fn Υ.Σ

Gx + y®nGi

Ex

y &lt; * &gt; nU

Ex. (21) i) Measure the displacement voltages C / 0 x and the sum currents! Χ_x according to known methods according to magnitude and angle for the mains frequencies per outgoing x at time U. The angle refers to the associated displacement voltage of the outgoing x. j) Measurement of the conductor ground voltages and calculation of the voltage L / i of the system at line frequency according to known methods. 6 AT 500 004 B1 k) calculation of Yu x per departure with: y. &Amp; (22) I) Divide the unbalance per departure on the three conductors: ϊι_χ = ϊζχ / 3 (23) (24) Ϊ2_χ = / 2ϊζχ / 3 = 3 ^, / 3 with Y ", = ^ - -1 - 2a2 \ Y - -Σ ^ S ~~ T ¥ ~ y2 = 2-y3 m) If the amount of deviation of the real part of Yj, Y2, or Y3 from the mean real part of Υχχ exceeds a set limit value at the exit x, a low-impedance error is detected , B) At time t2, a check is made for each output x for an additional high or low impedance unbalance or for a switching action in the outgoing circuit consisting of the following steps: a) Measuring the displacement voltages U0 x 2 and the sum currents _ * _ a according to known methods according to amount and angle for the grid frequencies per departure x. The angle refers to the associated displacement stress of the exit x. b) Calculate the additional asymmetry with yF = a t x t2 -Σ - Σ / 1 ^ -0 x t2 ~ - 0 x / 1 ^ .A | "2 ~ ~ - ^ - with / c = 1, a, a

Uq "t2 + KU, (25) c) Selection of that k at which the greatest active component in the fault admittance YF results, since an ohmic component is to be expected for the earth fault case. d) If YF exceeds a set limit value, method A) checks the sum admittance Vx. If the magnitude of the imaginary part of dYv = abs {imag {Y1 t2 - YXf1}} (26) exceeds an adjustable limit value, a switching action is detected and Y1 is adopted as the new reference value. In the other case, a high or low impedance ground fault is detected.

In the following the invention will be described by way of example with reference to the following diagram:

Figure 1 shows a deleted three-phase system with only one outlet. The power supply with a mixed stream consisting of up to two superimposed currents with different frequency and deviating from the mains frequency takes place in this example in the neutral point of the transformer. 7 AT 500 004 B1 For healthy network operation, the following equation can generally be written for the total current for the output x: u = (U.0 + UW) Y, _X HU0 + U2N) Y2_X + {U0 -3W) -3_x = ^ 0 (ϊι χ + ϊ2, + ϊ3, + ϊ2υ2_χ + §υ3_χ) (27)

The conductor-earth admittances normally consist only of an ohmic derivative and the conductor-earth capacitance. In deleted systems, however, distributed Petersen coils or single-phase compensation chokes are also possible. To take these into account too, the ladder-earth admittance is measured at two frequencies. As a result, a determination of the three components per departure is possible and the corresponding admittances can be calculated back to the network nominal frequency.

With the help of the power supply, it is possible to determine the sum admittance per outgo- ing at a frequency deviating from the mains frequency. Y * GX + j®nCEX + _l_ 'j (OnLEx ^ (28)

The measurement is not influenced by the often considerable disturbing influences in the line frequency range. For these non - line frequency measurements Um in equation (27) has to be set to zero. The measurement is possible by feeding in both symmetrical networks and during the earth fault. Switching operations are detected by changing the summation admittance. A suppression of overreactions of the relay in case of high-impedance errors is thus possible at any time and quickly.

By simultaneously evaluating the power-frequency components of (U0, and ίχ_χ, the imbalance in outgoing x can be determined.

Y U X &amp; (29)

The division of the asymmetry on the individual conductors can be carried out according to equation (23). This allows a complete description in the three-phase system.

If a high-impedance fault occurs, an additional admittance is connected in a conductor. In Fig. 1 this is represented by the switch S and the admittance yF. (30) with k = \ a2, a

The factor k depends on the conductor in which the ground fault occurs. By using the measured values at the time U and the summed admittance also at the time fi, an error impedance can be obtained by the following equation.

Yf =

(L Σ x (2 ~ ίτ_ 1 ^ · -Σ_π (-0 x t2 ~ -0 x n)

Uo, 2 + ^ 1 with / c = 1, a2, a (31) again selecting factor k depending on the assumed ground fault. Normally, a ground fault represents an ohmic resistance, so that k is chosen where the active part of yF is greatest. Simultaneously with this measurement, the current measurement of the sum admittance Υχ_Ά is possible. If an excessive change in its imaginary part is detected in comparison to the previously measured value ΥΣ_η, the deviation is not due to an additional imbalance but to a switching action. The associated ground fault message is suppressed.

However, the current asymmetry of the conductances Y | _x, Y ^ _x and Y3 x can be investigated. If the resistive component exceeds a defined value, this can be reported as a ground fault. The measurement and calculation of the distribution of Y | _x, Y2_x and Y3 x can be performed at any time with this new method, even during a low-resistance ground fault.

Due to the simultaneous measurement with three frequencies, the values are determined very quickly. The interference caused by fluctuating contact resistance is minimized.

Since the measurement of Υχ is not influenced by a load-dependent crosstalk of the line frequency system, a plausibility check can also be carried out with regard to load-dependent crosstalk and phase splitting in the meshed network. These effects led to uncontrollable false readings in the previous methods. Crosstalk effects are discussed in the article "Control of Petersen Coils, ISTET 2001". described in detail.

The previous methods also fail in networks in which the zero voltage or the zero current does not have constant values during the measuring period of up to a few minutes.

Further advantages of the method are: The measurement of the voltages for the frequency deviating from the mains frequency over the open triangle is much more accurate, since a summation of three large values, as occurs in the three-phase system, does not take place. - The measuring cycle essentially depends only on the filter algorithm used. - The measurement of Y ^ _x the departure can be done continuously. The measurement can also be made on request, e.g. after changing the power frequency shift voltage or at cyclic intervals, e.g. every 10 minutes. - The line capacity per departure is determined. These values can also be used to control a Petersen coil. - The measurement provides the Υχ_χ even for completely symmetrical networks. - Switching operations are also detected in symmetrical networks with this measuring method. - Fault detection of the earth fault detection can be quickly suppressed by the rapid measurement of Y ^ _x. - By quickly finding Y ^ _x, the actual location results are available much faster. This can take a few minutes with mains frequency power supplies if there are stationary conditions at the fault location. Even with smaller fluctuations, the previously known methods give no results whatsoever, since the determination of Ya does not yield any valid reproducible results. - Distributed Petersen coils in outlets can be detected and taken into account. For mains-frequency power feeds, current changes on compensated outgoing feeders are overstated.

Claims (16)

9 AT 500 004 B1 - The measurement is performed without any adjustment of the Petersen coil, which greatly reduces the mechanical stress on the Petersen coil. This can be very high, especially in networks with crosstalk of the load current to the zero system, since a calculation cycle is started with each load change. - During the measurement, the Petersen coil is not adjusted so that the required compensation is maintained for most measurement tasks. - The frequencies can be selected in a large detuning that is measured in the vicinity of the natural frequency of the network. As a result, an accurate measurement is achieved even for heavily detuned networks. 1. Method for detecting and locating low-resistance and high-impedance ground faults in electrical power supply networks characterized by the following steps: A) At time U, a measurement and calculation of the total energy to ground and a distribution of the unbalance to the individual conductors takes place at each outgoer Earth admittances consisting of the following steps: a) feeding a composite current into the null system of the deleted network, b) where the current from the superposition of the current with the amplitude / Λ and the frequency f, and the second current with the amplitude / ß and the frequency f2 composed; c) measuring the displacement voltages L / 0ji_x and U0_ & _ * according to known methods for the two frequencies per outlet x or on the busbar with clear assignment to the outlet x; d) measuring the sum currents Ιχ_η_χ and! xj2_x according to known methods according to magnitude and angle for the two frequencies per output x. The angle refers to the associated displacement voltage U0_fi_x or y0_f2_x of the outlet x; e) Determining the sum admittance per departure x for both frequencies: -Σ M x -Σ M x U or Y 0 M x -l_f2_x~ u-nx (D with + Y 2 fl X 1-3 f1 x γ = Y + Y -ς nxnj L · Gix + y®iGu + 7 jü>, L 1x- / G2x + ίω ^ 2χ +; ML; 2xJ G3x + 7 «> Ax + jü> f Ljx (2) 1 Gx + + - Y = γ + γ + γ i-Σ f2 x i-1 f2 x 1-2 f2 xi-3 f2 xr Gix + MC, x + 7 Gx + jo) 2CEx + jco2Lu) 1 G2x + jco2C2x + jo) 2L 2 xj G3x + j (02C3x + - jo> 2L3xJ (3) jco2LEx ^ f) the effective component Gx per output of the equivalent circuit results directly from the real part of the sum admittance; 1 0 Ϊιλ η = real ·, / 1 - G1x + G2x + G3x - Gx -0 H x AT 500 004 B1 (4) g) the calculation of the earth capacitance Cex and the inductance of any Petersen coil L &amp; on departure, each exit is made from the two imaginary parts by reshaping; Yii_n_x = -imag {Yz / n x} or Υυ f2 = -imag {ΥΣί f2 x} (5) GEx Σί_ί2_χω2 ~ ΥςΙ _ ^ _ χ0Χ \ 2 2. α> ι -ω2 Ι-Εχ ~ 1 ω · \ (Υΐί η χ + ωιgex) (6) (7) h) Calculation of the sum admittance per output for the line frequency fn Υς Gx + 1ωηΡεχ (8) i) Measurement of the displacement voltages Uo_x and the cumulative currents / i_x according to known methods according to magnitude and angle for the mains frequencies per outgox at time U. The angle relates to the associated displacement voltage of the outgoing x; j) measurement of the conductor ground voltages and calculation of the voltage L / i of the system at line frequency according to known methods; k) Calculation of Yu_x per departure with: -u-x V I) Break down the asymmetry per departure on the three leaders: (10) Ϊι_χβΪ * / 3 Ϊ2_χ = / 2ϊΐχ / 3 ^ 3_χ = ^ ΙΧ / 3 a-a 2 with (11) y = 2-y i-2 -3 where a corresponds to the known complex rotation operator .2 (12) a = e3 * = -0.5 + 0.866, / for a three-phase system; m) in the departure x exceeds the amount of deviation of the real part of Y ^, V2 or Y3 from the mean real part of Y ^ x a set limit, a low-ohmic error is detected. B) At time t2, a check is carried out for each output x on an additional high or 1 1 AT 500 004 B1 low impedance unbalance or on a switching operation in the outgoing consisting of the following steps: a) Measuring the displacement voltages U0_xj2 and the total currents! Χ2χ_α known methods according to amount and angle for the power frequencies per departure x. The angle refers to the associated displacement stress of the outlet x; b) Compute the additional imbalance with (U t2 "-Σ -Σ_Μ (-0 x / 2" -0 X / 1) U0 x t2 + kU, with / c = 1, er, a (13) c) Selection k, in which the greatest active component in the fault admittance YF results, since an ohmic component is to be expected for the earth fault case; d) If YF exceeds a set limit value, then method A) checks the sum-admittance Υχ. If the magnitude of the imaginary part of c (Ya = abs {imag {Yz, 2 - ΥΣ (1}} (14) exceeds an adjustable limit value, a switching action is detected and! Χ_χ_ \ ζ, Uo_x_t2 and Yij2 are adopted as the new reference value In other case, a high or low impedance ground fault is detected. 2. The method according to claim 1, characterized in that the power supply takes place in the neutral point of the feed transformer. 3. The method according to claim 1, characterized in that the power supply takes place in the neutral point of a zero point generator. 4. The method according to claim 1, characterized in that the coupling of the power supply via an auxiliary transformer or via a power assist winding of the Petersen coil takes place. 5. The method according to claim 1, characterized in that the section B) of claim 1 is repeated continuously. 6. The method according to claim 1 to 5, characterized in that the method is also applicable to other neutral point treatments or for two-phase networks or multi-phase networks. Method according to claim 1, characterized in that the determination of Υχ χ takes place continuously, i. a switching operation is recognized immediately and does not have to be requested via an additional calculation process. 8. The method according to claim 1, characterized in that in symmetrical networks, a continuous or a short repeated at defined time intervals feeding the mixed stream to detect switching operations in the departure as quickly as possible. 12 AT 500 004 B1 9. The method according to claim 1, characterized in that a change in the 50 Hz shift voltage trigger one or more measurement cycles for determining the summation admittance and the fault admittance per departure. 10. The method according to claim 1, characterized in that one or more measuring cycles for determining the summation admittance and the Fehleradmittanz per departure are triggered by an external signal. 11. The method according to claim 1, characterized in that the frequencies h and f2 are selected so that they are close to the resonant frequency of the network. 12. The method according to claim 1, characterized in that in noisy networks averaging over several measurements. 13. The method according to claim 1, characterized in that x is measured with missing compensation coils in the departure x only with a frequency f1. 14. The method according to claim 1, characterized in that by the comparison of Υχ_α with Υςμ interference by crosstalk of the load current to the summation current, for. be suppressed or compensated by a poorly balanced Holmgreenschal-tung. 15. The method according to claim 1, characterized in that caused by the comparison of Υχ_α with Y ^ _t1 interference caused by phase splitting in ring lines or compensated. 16. The method according to claim 1, characterized in that is regulated by specifying a maximum change in the displacement voltage of the feed current to an associated maximum value. For this purpose 1 sheet of drawings