Lab Report

Daniel Yoo ELEC4611 Christopher Lopes 1 of 15 LAB REPORT

ELEC4611: POWER SYSTEMS EQUIPMENT

EXPERIMENT 4: IMPULSE VOLTAGES IN ELECTRICAL SYSTEMS

BY DANIEL YOO (z3293531)

CHRISTOPHER LOPES (z3251736) Introduction An impulse is an aperiodic transient voltage or current which usually rises rapidly to a peak value and then falls more slowly to zero. Impulse transients can occur on electrical networks from direct hits by lightning strikes and switching operations. These instances can cause unpredictable voltages and currents that may be many times above the normal operating/rated limits. Therefore, it is essential to measure the potential effect of these events by impulse testing. [1],[2] A distinction is made between lightning and switching impulses on the basis of duration of the front. Impulses with front duration up to 20µs are defined as lightning impulses and those with longer fronts are defined as switching impulses. [1] In part A of this experiment we examine the effect of lightning strikes on HV equipment. This is done by using a four stage impulse generator to produce a 1.2/50µs waveform. We aim to determine at what voltage level causes breakdown across a sphere gap. The term “electrical breakdown” relates to phenomena associated with the failure of insulation under electrical stress, in which the discharge completely bridges the insulation under test, reducing the voltage between the electrodes practically to zero. The term “flashover” is used when a disruptive discharge occurs over the surface of a dielectric in a gaseous or liquid medium. [1], [2] In part B of this experiment we examine the distribution of voltages along different types of transformer windings. A 0.2/150µs waveform is used. A comparison is made between conventional windings and interleaved windings. [3]

Part A: High voltage impulse testing Introduction Much of equipment on a power system is subjected to surges due to lightning strokes. Each system is designed for a basic insulation level (B.I.L.) requiring the various components to be capable of withstanding different values of impulse voltage depending on the insulation coordination philosophy adopted. The shape of impulse voltage chosen to simulate the lightning conditions is 1.2/50µs (see AS1931) wave: the voltage magnitudes vary according to the equipment and are specified in the particular apparatus specifications. The objectives of the experiment are:

i) Identify various components of a four‐stage impulse generator and d.c charging set ii) Become familiar with two methods of HV measurement (sphere gap and resistive

divider) and make a direct comparison at high voltage iii) Determine operating characteristics of a surge diverter, used to protect equipment

against lightning. [3]


Figure 1.Waveform of a model lightning strike [2]

Background Theory To test equipment in this experiment we use a four‐stage impulse generator to provide a waveform that models a lightning strike in accordance with AS1931. The impulse generator is D.C. supplied from a rectifier and a transformer. This charges each capacitor in the four stages. When the spark gap in the lowest stage is triggered or the voltage exceeds the breakdown voltage it flashes over. This raises the voltage seen by the second stage and places twice the single initial D.C. voltage over the second stage spark gap which causes it to flash over. This in turn raises the third stage voltage which in turn raises the fourth stage voltage. In this fashion four times the initial D.C. voltage is seen at the output. The output waveform rise can be controlled by setting the rise time constant which is determined by the front resistors and front capacitor. The fall time is similarly determined by the tail resistors and generator capacitors. A resistive voltage divider allows a CRO to connect to the output of the impulse generator. A surge

diverter is placed in parallel with the CRO connection as a malfunctioning earth could result in

destructive voltage levels at the CRO terminals. [3]

Apparatus The standard lightning impulse voltage is supplied by a Passoni & Villa 400kV/10kJ impulse generator. The circuit is shown in figure 2. It has 4 stages in series. The charging volts per stage is 100kV and the generator capacitance per stage is 1µF (two 0.5µF capacitors in parallel). Thus the rated energy of the generator is 10kJ.

Resistive divider (200:1)

Digital storage oscilloscope

Sphere gap (25cm in diameter) for calibration

24kV surge diverter (test piece)


Figure 2. Set up of HV apparatus for impulse testing

Procedure

1. Verify that the circuit gives a 1.2/50µs wave by recording the full wave on CRO. Compare the measurement results with the calculated values using the formulae:

2.75 and 0.70

Where is the front capacitance, the front resistance, the generator capacitance and

the tail resistance. Note that both the external and internal front resistors should be included in the calculations for .

2.75

Where 4342 and 87Ω 55Ω 4 8Ω

2.75 4342 87Ω

1.04μ

0.70

Where 0.25μ 4 1μ and 320Ω 4 80Ω

0.70 0.25μ 320Ω 56.00μ


Figure 3. Rise time of test waveform Figure 4. Fall time of test waveform

From figures 3 & 4 above we can clearly see the rise and fall times. The rise time can be seen on the left as the time to rise to the peak value. The fall time can be seen on the right as the time to fall to half the peak value. They are as follows: 1.4μ

52.4μ

AS1931 Values Calculated Measured

1.2µs±30% 1.04µs 1.4µs

50.0µs±20% 56.0µs 52.4µs Table 1. Verification of test wave to AS1931

The calculated value for rise time is 12% below the standard value whereas the measured value is 16% above. The calculated value for fall time is 12% above the standard 50.0µs whereas the measured value is less that 5% different. This is within the limits set out by the AS1931 and is suitable for this experiment. [2]

2. Calibrate the sphere gap set for spark‐over at 70kV. Lower the impulse level as far as possible to achieve at least five chopped waves for 10 successive impulse applications.

The estimated value for flashover can be determined by finding the minimum such value that causes at least five waves to be chopped out of ten applied waves. This occurred when the CRO read 15.2V as can be seen in figure 5 below:

Figure 5. Chopped wave


3. Open the sphere gap and increase the impulse level until the surge diverter begins to chop. Observe the waveshape and record the residual voltage level.

Figure 6. Surge Diverter begins to conduct

In figure 6 above, the surge diverter is beginning to chop the waveform. This is seen by the tail, which falls steeply at first and then more gradually. This indicates two different time constants caused first when the surge diverter is on, bypassing the tail resistors, and then when it returns to being open. The residual voltage is therefore the voltage at which the surge diverter returns to an open circuit. This occurs on the CRO at approximately 13V which equates to: 204.3 24 13 63.7

Results and Discussion

1. Compare sphere gap flashover value with that estimated from the divider and CRO

The nominal value for the sphere gap flashover is 70kV. The voltage divider ratio is 203.4:1 and the voltage transducer a further 24:1. Therefore, the actual flash over voltage is given by: . . 203.4 24 15.2 74.2 This is an error of 6% which can easily be accounted by changes in humidity affecting the breakdown of air, the error in separation of the sphere which is manually controlled and the accuracy of measuring equipment.

2. Comment on the time lag of the surge diverter flashover

The time lag between the application of the surge and the flash‐over of the surge gap arises due to the charging time as determined by the front capacitor and the external front resistance. This is expressed in the given equation: 2.75

3. Describe the principle of the controlled tripping (triggering) circuitry

The controlled tripping circuitry works by attempting to induce breakdown across the lowest spark gap at a voltage determined by the operator. Without this mechanism, flash‐over will occur only when the voltage is large enough to overcome the spark gap and this voltage can vary from test to


test. The trigger gives the operator a way to ensure the flash‐occurs at a consistent voltage each time and meaning the output impulse is the same each time which increases the reliability, repeatability and credibility of the experiment. Importantly, the voltage determined by the operator must be above a certain level or else even the trigger will fail to instigate flash‐over. Once the lowest spark gap has flashed over, there is a virtual short to the negative terminal of the next capacitor. This has some voltage across it which increases the voltage seen by the next highest spark gap and leads to subsequent flash‐over. In this fashion, each stage contributes as the spark gaps progressively accumulate the capacitor voltage to produce the impulse. Triggering the lowest spark gap at a consistent voltage allows the output impulse to also be consistent in amplitude and time.

4. If the object for the lightning impulse test is a length of HV power cable which has a capacitance of 5nF, what adjustments to the circuit of Fig. 1 are required?

The HV cable would act as the test capacitor and so it should replace the front capacitor. As the capacitance of the cable is slightly higher than the front capacitor it will also be necessary to adjust the front resistors to achieve the same rise and tail times as set out by AS1931.


Part B: Voltage distribution along transformer windings when subjected to a unidirectional surge

Introduction

All power systems equipment is designed to withstand the surges produced by lightning strikes and switching operations which occur on the overhead lines. The most expensive items subject to such events are the large power transformers located at power stations and sub‐stations. Large rotating machines are also susceptible to such surges because of their winding structures. [3], [4] The present experiment is aimed at demonstrating the manner in which a lightning surge is distributed within two types of transformer windings. The technique used is still employed to determine such distributions in practice. The measurement highlights some of the difficulties encountered in design and in simulating full‐scale effects. The possible errors in choosing data for computer programs are apparent, e.g. in the representation of distributed parameters of inductance and capacitance. [3], [4]

Background Theory A transformer winding can be simplistically represented by a uniformly distributed circuit of series inductance and shunt/series capacitance. When a unit‐function voltage is applied to this equivalent circuit, the initial distribution of voltage is determined entirely by the capacitive network shown in figure 7. The voltage at a point distant from the neutral end of the windings is given by:

Where

distance from the neutral end

Figure 7. Representation of a transformer winding at transient frequency


The final voltage distribution is determined by the series resistance, or if there is a superimposed power frequency excitation, by inductance; the distinction is superfluous since both quantities are usually linear, and the distribution therefore becomes a straight line, corresponding to α= 0.

Figure 8. Voltage distribution along transformer windings

when subjected to a unidirectional surge We can see in Figure 8 that alpha can relate to how much voltage distribution is concentrated in the line. In order to make the most efficient use of the insulation provided in the transformer, it is imperative to design the windings in such a way as to eliminate, or at least reduce, both the initial voltage concentration at the line end, and also subsequent high oscillatory voltages in the body of the winding. [4] Most of this report will look into how we decrease the earth capacitance and increase series capacitance to get α as close to zero as possible. Two possible methods in improving the voltage distribution are:

1) Interleaved winding 2) Use of stress grading rings [4]


Figure 9. Section through a) conventional disk,

b) disc with 2 turns in parallel and c) interleaved disc

Figure 10. Capacitance network of disc type coils for a) simple conventional disc and

b) interleaved disc.

Apparatus

Low‐voltage pulse generator

Digital storage CRO

66kV, 6MVA (one phase) conventional disc winding and low voltage windings

66kV, 25MVA (part winding) interleaved disc, 3 conductors in parallel, no low voltage winding, inner earthed screen

Procedure A. 66/11kV, 6MVA conventional disc winding

1. Note layout of windings – number of disc and turns, connections, etc. See Figures 9(a) and 10(a) of Background Theory. Also note low voltage connections.


2. Connect output of pulse generator to HV and the winding neutral to ground. Note earth connections.

Figure 11. circuit used to measure winding voltage distribution

3. Set pulse generator to give an impulse of approximately 0.2μs rise time and a small decay

over a time of 150μs. 4. Measure and record waveshapes (magnitude and duration) at crossovers between discs.

Every third connection will suffice except where finer detail is required.

Figure 12. Waveform of incident voltage (yellow) and

measured down the line voltage (purple)

5. During (4) attempt to identify “travelling wave” component and record apparent magnitude and “time delay”. The delay appears to be about 10μs for the complete winding. Winding #

(n) Vpulse (V)

Vpeak (V)

V ratio (%)

t_delay (us)

Distance to neutral (%)

3 8.25 7.45 90.30 0.6 91.67 6 8.16 6.50 79.65 1.2 83.33 9 8.37 4.29 51.25 2.0 75.00 12 8.20 3.54 43.17 2.9 66.67 15 8.25 3.20 38.78 3.5 58.33 18 8.29 2.29 27.62 4.5 50.00 21 8.25 1.79 21.69 4.5 41.67 24 7.58 1.37 18.07 6.6 33.33 27 8.25 1.41 17.09 7.8 25.00 30 8.00 1.04 13.00 8.3 16.67 33 8.12 0.50 6.15 8.9 8.33

Table 2. Conventional winding measurements


B. 66kV, 25MVA interleaved disc coil stack (no LV winding)

1. Repeat items A1 to A6. Is there a “time delay”? Refer to Figs. 9(b), 9(c) and 10(b) of the Supplement. Determine conductor layout (three in parallel).

Figure 13. Waveform of incident voltage (yellow) and

measured down the line voltage (purple) Interleaved winding achieves a capacitance increase and a leakage inductance decrease which reduces the ‘time delay’ making it difficult to measure.

Winding # (n)

Vpulse (V)

Vpeak (V)

V ratio (%)


3 8.33 6.70 80.43 85.71 6 8.12 5.00 61.57 71.42 9 8.08 3.66 45.29 57.1412 8.25 2.62 31.75 42.8515 8.00 1.41 17.62 28.57 18 8.08 0.46 5.69 14.28

Table 3. Interleaved winding measurements

Figure 14. Voltage distribution of conventional winding

Alpha Conventional Winding

0

20

40

60

80

100

0102030405060708090100


Vo

ltag

e to

GN

D (

%)


Figure 15. Voltage distribution of interleaved winding

In this experiment we measured the effectiveness of the interleaving method to improve voltage distribution. α is a function of series capacitance and earth capacitance. We can get this closer to zero by decreasing the earth capacitance or increasing the series capacitance. This is exactly what the interleaved winding does; it will increase the series capacitance to make the voltage distribution more linear across the line. It is evident that the voltage distribution is more evenly distributed along the line of the interleaved winding in Figure 15. This graph shows that with effective interleaved windings, we can get α to be much closer to zero.

Results and Discussion

1. Plot “initial” voltage (<1us) magnitudes against position in winding for cases A & B.

Figure 16. Initial Voltage vs Winding Position of Conventional and Interleaved winding

Alpha Interleaved Winding

0

20

40

60

80

100

0102030405060708090100


Vo

ltag

e to

GN

D (

%)

Initial Voltage vs Winding Position

7.5

7.6

7.7

7.8

7.9

8

8.1

8.2

8.3

8.4

8.5

0 5 10 15 20 25 30 35

Winding Position (n)

Vo

ltag

e (V

)

Conventional

Interleaved


2. Estimate value of 'alpha' from graphs in the Supplement. From Figure 14 we estimate that: α = 3 for our measured conventional winding. The voltage distribution is more concentrated at the line end, causing higher stress across those windings and their insulation. The higher grading of α refers to higher concentration of voltage distribution at one end. For example if α = 30 there is 90% of the voltage distributed on the first 10% of the transformer windings. From Figure 15 we can see that α has been significantly improved and for the interleaved winding it is approximately: α = 1

3. For case A plot (i) apparent time delay versus winding position and (ii) magnitude at delay time versus winding position.

Figure 17. Time delay vs. Winding Position

Figure 18. Magnitude at time delay vs. Winding Position

Time delay vs. Winding Position

0

1

2

3

4

5

6

7

8

9

10

0 5 10 15 20 25 30 35


Tim

e d

elay

(u

s)

Magnitude at time delay vs. Winding Position

0

1

2

3

4

5

6

7

8

0 5 10 15 20 25 30 35


Vo

ltag

e (V

)


It is evident that as time delay increases, the voltage decreases. This is only for the conventional winding as the time delay causes the magnitude peak to occur after the peak of the incident voltage.

Discussion

1. Reason for low voltage (and equivalent high voltage) tests. Power system equipment are placed in a variety of environments, with some being susceptible to lightning strikes and other forms of overvoltage from switching, etc. A test capable of simulating these overvoltages ensures that the equipment is capable to be situated in such environments and not be damaged and cause outages, loss of property, or personal injury. Sometimes equivalent low voltage tests are used (part 2 of this experiment) because the equipment to produce high voltages is costly, large and dangerous. In this case the low voltage tests was purely to show us the voltage distribution across windings. It lacks in the area of testing the Basic Insulation Level (BIL) level or quality of the insulation as a HV surge may have caused insulation breakdown.

2. Neutral current "fault" detection. This technique monitors the neutral currents during impulse testing. It helps us detect breakdown of insulation that may occur during HV winding tests as the wavelet at the neutral point may be distorted due to an inter‐turn short circuit.

3. How interleaved winding improves HV impulse voltage performance. Alpha is a function of series capacitance and earth capacitance. We can get this to 1 by decreasing earth capacitance or increasing the series capacitance. This is exactly what the interleaved winding does; it will increase the series capacitance to make the voltage distribution more linear across the line, reducing the chance of insulation breakdown occurring across line end windings.

4. Why there is no clear travelling wave component in case B. It could be due to the increase of series capacitance on the line, because the added capacitance increases the impedance. Also the interleaved winding technique reduces the leakage inductance which reduces the time delay. All this makes the measurement of the time delay more difficult with the appearance of a no clear travelling wave.

5. The winding is normally immersed in insulating oil of εr ≈ 2.2. What effect would this have on the results if they were done with oil?

2

Introducing an εr will increase all the capacitances. The increase should all be in proportion so that the voltage distribution remains unaffected. Two main effects come from immersing the windings in oil. Firstly, the breakdown voltage will increase, thus reducing the chance of partial discharge or arcing occurring during a lightning impulse. Secondly, the oil would aid in heat convection and will therefore increase either the lifetime of the equipment or the rating of the equipment.


6. Switching surge voltage waveform has slower rise and fall times, e.g. 100/1000μs. What effect would this have on voltage distribution on windings?

Switching surges occur as a result of a circuit breaker or switch breaking a fault current. Switching surges last longer than lightning surges because they contain more energy. By lasting longer they contain lower frequency components than lightning surges, therefore reducing the dominant capacitive attribute of the windings. The switching surge should have a more uniformly distributed voltage across the transformer windings.

Conclusion In this experiment we performed an impulse test to simulate a lightning strike and observed the

distribution of voltage over different types of transformer windings under transients.

In part A, we produced a test waveform from an impulse generator that complied with AS1931 by

setting the parameters of rise and fall times. The various components involved in the four‐stage

impulse generator include generating capacitors, spark gaps and resistors which form a stage. The

D.C. charging set includes a step‐up transformer and rectifier unit.

A lot of different environmental factors and human errors impacted the test results in this

arrangement which means that calibration is needed every time. This is not optimal, as the

capacitance of the test object also changes the parameters of the waveform and so the resistors and

capacitor values would also need to be changed to keep the waveform in compliance with AS1931.

In this experiment we observed that the surge diverter exhibited a time delay before it conducted

and that the residual voltage was approximately 60kV.

In part B, we observed the voltage distribution across a conventional transformer winding and an

interleaved one after a transient was applied. The conventional winding exhibited a non‐linear

voltage distribution with high stress at the line end and low stress at the neutral end. In comparison,

the interleaved winding had a much more linear voltage distribution which means that the stress is

more even along the length of the winding. This occurred because interleaved windings increase the

series capacitance and hence send α closer to unity.

References [1] Australian Standard AS1768:2007 Lightning Protection [2] Australian Standard AS1931:1996 High‐Voltage test techniques [3] Dr. Toan Phung, ELEC4611 Lecture Notes 2012, Electrical Insulation for High Voltage Equipment, University of New South Wales [4] Dr. Toan Phung, ELEC4611 Experiment 4, Impulse Voltages in Electrical Systems, University of

New South Wales