Israel electric co. has begun research and development of patch antenna use as a very sensitive mobile window ultrahigh-frequency (UHF) sensor to detect partial discharges (PDs) in gas-insulated substation (GIS) equipment.

THE PROBLEM

The UHF technique has proved to be an effective diagnostic tool for measuring the PD when testing the integrity of the dielectric strength of GIS. The GIS chamber acts as a coaxial cylindrical resonant cavity in which the PDs excite electromagnetic waves, which creates resonance above and beyond the cutoff frequencies' transverse electric (TE) and transverse magnetic (TM) modes. These resonances are detected as PD signals at UHF frequencies (300 MHz to 3 GHz) by built-in plate sensors that are similar in design to capacitive couplers. The propagating signals are damped due to reflections at barriers and other discontinuities, dispersion or division at tee junctions, and attenuation due to power dissipation in the metallic enclosure.

As a consequence of the UHF damping, many preinstalled internal sensors have to be installed in an extended GIS. For new or refurbished GIS, these internal sensors are manufactured and fitted by the switchgear manufacturer. For old versions of operational GIS, new sensor installation requires circuit outages and dismantling of the enclosure. As this work is not always justified, the only alternative for these GIS installations is the use of external sensors and various types of these are in use.

For certain GIS, the barrier insulator includes potential grading electrodes, which could be used as a UHF sensor. An external antenna also can be used as a UHF sensor, which is fitted to the barrier insulator that has no metal case. Almost all the early GIS disconnector and earthing switch installations have inspection windows that offer the opportunity for optical inspection of the isolating distance in the disconnectors, so insulation monitoring can be performed using mobile sensors fitted to these windows. The disadvantage of mobile sensors is their low sensitivity compared to internal sensors. The window construction works as a high-pass filter so that, below cutoff frequency, the signals are significantly damped. Only for large-window diameters, short lengths and high measuring frequencies (more than 1.3 GHz) can the sensitivity of mobile sensors be compared to those of internal sensors.

THE WORK

Israel Electric Co. (IEC) began research and development (R&D) on sections of GIS at different SF₆ pressures to examine the following aspects of the problem:

• The influence of external disturbances on patch antenna measuring results

• Sensitivity of the method at different SF₆ gas pressures of GIS

• Influence of the distance between the sensor and the PD source

• Influence of the type of PD source

• Practical use of the patch antenna as a UHF window sensor for detecting PD signals in operational GIS.

R&D TEST PROCEDURE

The utility performed tests on one phase of a GIS busbar section rated at 170 kV. This module with aluminum-welded enclosure included four SF₆ compartments separated by barriers. Compartments No. 1 and No. 4 were corners, and compartments No. 2 and No. 3 were tubes 1 m and 2 m (3.3 ft and 6.5 ft) in length. An air/SF₆ bushing was connected to corner compartment No. 4, and the flange of corner compartment No. 1 had an inspection window to which the external antenna was fitted. The module is shown in Fig. 1.

IEC conducted tests at the following range of SF₆ relative pressures: 0 bar, 1 bar, 1.5 bar, 2 bar, 3.5 bar and 5 bar (0 kPa, 100 kPa, 150 kPa, 200 kPa, 350 kPa and 500 kPa).

The test module was energized via a high-voltage transformer having a PD-free air connection to the bushing. UHF signals inside the test compartments were excited by a metal needle that was positioned in three different places: in the enclosure at points A and B (Fig. 1) and on the conductor in compartment No. 1.

External disturbances such as outside corona were simulated by exciting the corona discharge in air from the bushing shield.

The charge injected by internal corona when a needle was positioned on the enclosure was measured directly. For this purpose, the needle was insulated from the enclosure by a thin insulated layer and was connected to earth via an integrating capacitor. For the conventional PD measurements, based on International Electrotechnical Commission standard-60270, a 1-nF coupling capacitor with an active coupling quadripole connected to a standard PD detector (20 kHz to 500 kHz) was used providing a reference with a noise level less than 1 pico-Coulomb (pC).

The UHF PD signals from the patch antenna were picked up by an HP 8560 E (30 Hz to 2.98 GHz) spectrum analyzer connected to the antenna via a standard Mini-Circuits HF preamplifier (gain 20 dB). The analyzer was tuned to frequency range 700 MHz to 2000 MHz, while specific frequencies were investigated and displayed against the phase angle of the applied voltage to show the discharge point-on-wave (POW) records. All the records were performed at PD inception voltage. A triple-layer triangular patch antenna was used (Fig. 2).

Patch antennas became very popular in the 1970s because of their exceptional advantages of low weight, design simplicity and low cost. Their major disadvantages back then were narrow bandwidth and low gain. Relatively wide bandwidth and fair gain were reached using patches whose sizes decrease gradually from the bottom to the top layer. The antenna included three substrate-patch pairs, each containing a different-sized triangular patch separated by substrates of the same thickness and permittivity. The antenna had a bandwidth of about 570 MHz around the central frequency 1.95 GHz, and an average gain 7.3 decibels with respect to an isotropic antenna dB isotropic.

RESULTS

Figure 3 presents the UHF PD spectrum and POW record for the selected 1.53-GHz frequency obtained for an SF₆-filled module under 5 bars (500 kPa). It could be seen that the spectrum has significant signal energy in the frequency range between 1.3 GHz to 2 GHz. The POW record shows the excitation of the bipolar discharge that has different characteristics at negative and positive half waves. The reference UHF PD spectrum at zero voltage is shown in Fig. 3c.

Figure 4 shows the results obtained for the SF₆ pressure of 3.5 bars (350 kPa). The UHF spectrum and zero span recording are similar to those obtained for the module with 5-bar (500-kPa) SF₆ pressure. The conventional PD pattern for this case is shown in Fig. 4c. The different discharge intensity for positive and negative half waves with maximum PD level 7.5 pC is clearly seen on this PD pattern.

No remarkable changes were obtained when the SF₆ pressure was reduced to 2 bar (200 kPa) as shown in Fig. 5. However, when the SF₆ pressure was reduced to 1.5 bar (150 kPa), additional peaks appear on UHF spectrum at the frequencies below 1.3 GHz as shown in Fig. 6. This also shows the growth of signal energy in the region below 1.3 GHz at SF₆ pressure of 1 bar (100 kPa).

When the SF₆ pressure was reduced to atmospheric level, the inception voltage and inception discharge activity decreased (Fig. 7). This can be seen on POW recording (for frequency 1.67 GHz) and on PD pattern. At the same time, the POW records still clearly show the inception of the discharge on negative and positive half waves. On the PD pattern recorded with a relatively high sensitivity (1 pC) only unipolar discharge could be reliably detected. The UHF measuring method used in this case seems to be more sensitive than the conventional method.

Direct measurements of the injected charge from the needle for this case also show the discharge excitation at the positive and negative half cycles (Fig. 8). This UHF measuring method has shown high sensitivity up to maximum examined distance between the antenna and the PD source of 2 m (6.5 ft). The results show that the use of patch antenna as a window UHF sensor provides highly sensitive UHF PD detection on SF₆ equipment.

VERIFICATION

Verification of patch antenna use was performed on a 400-kV operational GIS installed in 1990. The GIS is equipped with a large number of inspection windows for disconnectors that facilitated performance verification by the injection the UHF signals through one window and reception of the signal by a patch antenna fitted to the other window. The window structure is shown in Fig. 9.

The response signals of patch antenna were measured by injecting the signals from the standard UHF pulse generator using a special directed transition antenna. The generator was able to produce a series of identical pulses for simulating the PD activity. The UHF spectrum measured for different output voltage of the generator is shown in Fig. 10 together with reference spectrum detected at zero output voltage of the generator. The verification signals are clearly detected in the frequency range 1.3 GHz to 1.5 GHz. The results obtained demonstrate the practical possibility of using the patch antenna for PD detection in GIS.

IEC's R&D work confirmed that the patch antenna could be used as an external UHF PD sensor for GIS applications. IEC also concluded that external noise such as corona discharge had no affect on the output of the patch antenna. These laboratory-based experiments demonstrate that the application of a patch antenna as an external UHF PD sensor has a high-sensitivity characteristic to affect detection. Furthermore, the research confirms that a decrease in the SF₆ gas pressure results in a spread of the UHF spectrum to the low frequency region.

The value of the patch antenna when used as a UHF window sensor was confirmed by the verification on operational GIS. The encouraging results have established a firm basis for further IEC work to develop new patch antennas with enhanced performance.

---

Vladimir Brandenbursky worked as a scientist in the High-Voltage (HV) Laboratory at the Moscow Institute of Energy on the development of new methods of investigation of discharge phenomena and HV test techniques. In 1984, he was awarded his Ph.D. from the Leningrad Polytechnic Institute. Since 1991, he has worked in IEC's Central Electrical Laboratory in the field of on-site HV testing and development of new diagnostic techniques for GIS. vbrand@iec.co.il

Dmitry Shevchenko joined the Central Electrical Laboratory of IEC as a test engineer in 1998 after graduating with a MSEE from the Krasnoyarsk Technical University. He is responsible for the research test arrangement and computer support.

Danny Shukrun joined the Israel Electric Co. in 1992 and since 1995 has worked in the Operations department. In 2003, Shukrun earned a bachelor's degree from the Coventry University branch in Roopin College, and in 2004, he finished his studies in high-voltage technology at the Holon Academic Institute of Technology.

Gady Golan is a professor and the department head of Electrical and Electronics Engineering at the Holon Academic Institute of Technology in Israel, where he is responsible for heading the Microelectronics and Thin Films Laboratory. He also serves as chairman of the IEE Israel and IEEE-EDS Israel.