K.P.A. N. PATHIRANA, A. D. RAJAPAKSE, University of Manitoba
R. WACHAL, Manitoba HVDC Research Centre, CanadaAccurate location of the faults in HVDC transmission lines, which are employed to transport large amounts of power, is essential for taking corrective measures quickly and cost effectively. Travelling wave based fault location has been very successfully used for line fault location in line commutated converter (LCC) based HVDC schemes. Travelling wave based fault location systems give good accuracy, even for very long transmission lines [1]. However, there is little experience in application of these fault locaters in new voltage source converter (VSC) based HVDC systems. Many of the proposed VSC HVDC applications such as interconnection of offshore wind farms involve submarine or underground cables. Fault location in cables is more challenging than in overhead transmission lines, and at the same time demands higher accuracy of location. VSC HVDC schemes with long cables, over 300 km , have been proposed and therefore, it is very important to adapt the travelling waves based fault location technology for VSC HVDC systems and improve the accuracy to deal with extreme cases such as very long cables.
Although calculations involved in travelling wave based fault location schemes are simple in theory, their implementation is challenging due to various factors that contribute to errors. These include bandwidth limitations of transducers, A/D conversion and sampling precision, synchronization errors, wave front detection algorithm errors, propagation velocity deviations due to changes in physical parameters, and the propagation velocity variations in different frequency components of the travelling wave in lossy transmission lines. To increase the accuracy of the fault location scheme, improvement of every factor mentioned above is necessary. In this paper, we focus on the effect of propagation velocity variations in different frequency components of the travelling wave.
In a transmission line, high frequency components of a fault generated travelling wave travel at faster velocities than low frequency components. Fault generated transients contain a range of frequency components extending from low frequencies to several hundred kilohertz. High frequency components of the travelling wave, although travel faster, subjected to more attenuation as they travel along a cable, mainly due to the high dielectric losses [2]. Furthermore energy of the travelling wave is reduced due to I2 R losses along the cable. If a fault happen closer to one end, the travelling wave that propagate over a longer distance get attenuated, and it may not contain the highest frequency components when it reaches the other end. On the other hand, the travelling wave arriving at the closer
end would contain almost all frequencies. Errors occur when the travel time difference is estimated
using two signals with different frequency contents. Although, this error is not significant in overhead
lines and short cable, ignoring these errors due to uneven wave front attenuation leads to unacceptable
errors in long cables.
This paper investigates the variation of the shape of the fault generated travelling wave and its
effects on the speed of the travelling wave based fault location through simulations performed in
PSCAD/EMTDC. The paper proposes a Rogowski coil based transient measurement system and
simple filtering scheme to ensure that signals with the same frequency contents are considered for
travel time difference estimation. Accuracy of the fault location calculation after the modification is
confirmed through simulations.
