The length of XLPE-insulated underground cable lines in the European HV / EHV network is experiencing a fast growth due to the increasing number of submarine interconnections and suitability to be installed also in urban area with nil visual impact after installation. Use of cable stretches in combination with overhead lines has proven to be a suitable means to ensure feasibility of new EHV transmission circuits. The design of long cable systems to be installed in HV and EHV networks is often associated with detailed transient and harmonic studies, in order to identify potential network problem areas.
Accurate and reliable models are needed to this purpose: time-domain simulation of power cables in the frequency range of interest to power system transients, i.e. from dc to some tens kHz still poses difficulties, especially when involving metallic sheaths. On the other hand, frequency domain calculation of admittance and impedance matrices for underground cable systems is nowadays well established, for any cable arrangement of interest for power transmission applications.
Modal analysis techniques with constant transformation matrices, successfully applied to overhead lines, are less suited to underground cable systems on account of the strong frequency dependence of transformation matrices. Other models accounting for frequency dependence of transformation matrices and based on fitting techniques may suffer numerical instability due to computation (truncation) errors, unstable poles fitting of the transfer matrix and interpolation errors on modal time delays that are not in general integer multiples of the simulation time step in EMTP-like software.
In order to estimate the accuracy of the current available models, TERNA and CESI carried out a test campaign aimed at time and frequency domain measurements on HV underground cables (MisCavAT); the field tests have been carried out on a 400 kV underground line (UGL) manufactured by Prysmian, laid in flat formation between Fontelupo and Tavarnuzze (Florence, Italy).
High frequency field tests on UGLs have been rarely performed, often evidencing a discrepancy between time and frequency domain measurements and simulation results; these discrepancies have also been associated by insufficient accuracy in ground return impedance.
The paper focuses on the results of the second MisCavAT test campaign which is part of a more comprehensive testing program foreseen in the Italian HV transmission network.
Tests have been carried out using three phase variable-frequency current injection, at positive and zero sequence, on both core conductors and metallic sheath.
Time domain tests, using mercury relays to cause switching transients, have been performed for validating the frequency-domain measurements. Results show a substantial agreement between the two techniques.