Modellbasierte Bestimmung der SF6-Verlustrate in Gasisolierten Schaltanlagen (English shop)

Preview

Table of Contents, Datei (50 KB)
Extract, Datei (1000 KB)

The IEC (International Electrotechnical Commission) standard 62271-203 permits a maximum SF6 leakage rate of 0.5% per year in gas-insulated high-voltage installations. This gives rise to the need to be able to measure, or even permanently monitor, such small leakage rates. To date, their detection has not been solved satisfactorily. Problematic are, on the one hand, natural, thermally induced fluctuations in pressure and local density values and, on the other hand, the high demands placed on the measurement accuracy and long-term stability of the sensors to be employed.

The fluctuations mentioned result from an inhomogeneous, transient temperature field in the gas compartment, which is caused by the heating of certain components due to ohmic losses as well as by time-varying ambient conditions such as outside temperature and solar radiation. These fluctuations can readily exceed the maximum leakage rate permitted by the IEC by a factor of ten. Leaks may thus remain concealed over extended periods.

Until now, the fluctuation of the measurement signal has been countered by subsequently smoothing it, for example by forming the moving average. This makes it possible to damp the diurnal variations. Seasonal variations, however, can hardly be reduced in this way. Assuming that the fluctuations are deterministic in nature, it was postulated that they can be calculated. To this end, an investigation of those phenomena that cause the fluctuations became necessary within the scope of this work. Numerical flow simulations (CFD, Computational Fluid Dynamics) and the experimental measurement of the flow field allowed the quantification of the transient convective flow in the circular cylindrical annular gap, as it occurs approximately, for example, in gas-insulated lines (GIL) and busbars of gas-insulated switchgear (GIS). It was shown that the strongly heated inner cylinder – that is, for example, the conductor tube – causes a pronounced convective flow of up to 7 cm/s above the inner cylinder. However, this flow leads only partially to the mixing of warm and cold gas zones. In particular, a zone of relatively cool and thus dense gas forms beneath the inner cylinder. If a density sensor were deployed at this location, comparatively large fluctuations would have to be expected.

In order to be able to measure the time histories of gas pressure and gas density at various positions, a test rig with a GIS busbar segment was set up in the laboratory. Various current load curves, combined with solar radiation, could be prescribed. It was shown that the pressure fluctuations are approximately three times larger than the highest density fluctuations (11% compared with 4%). The uniform pressure distribution means that the pressure fluctuations have the same value everywhere in the gas compartment. The inhomogeneous density distribution, by contrast, leads to significant position-dependent differences in density in terms of the shape and amplitude of the fluctuations. Maximum density fluctuations occur at the lowest point in the gas compartment, minimum density fluctuations at mid-height. As could be demonstrated on the test rig, this also applies to vertical elements such as those occurring in GIS and SF6 bushings.

A model-based approach is proposed as a possible filtering method for reducing the fluctuations. Here, the measured value is permanently compared with a simulated value. The deviation then corresponds to the gas losses due to a leakage. In order to keep the complexity, and thus the computational effort, of this model low, gas pressure is recommended as the measurand. Unlike gas density, it is homogeneously distributed. In this way, it is realistic to be able to detect a real gas loss of 0.5%. Under laboratory conditions, a model accuracy of even 0.3% could be achieved, which correspondingly reduces the response threshold for leakages to 0.3%. This presupposes a correspondingly high measurement accuracy and long-term stability of the pressure sensor. Gratifyingly, a suitable and readily available pressure sensor could be found.

The present work shows that the detection of gas leaks in accordance with the IEC specification of a maximum of 0.5% per year is demanding and, to date, had not been solved satisfactorily. For the first time, the model-based filtering method presented here provides a procedure which, in combination with a carefully selected sensor and, where applicable, its optimal positioning, can measure such small leakages and thereby detect an impermissibly high leakage rate.