Abstract
In attempting to analyze various experimental data that have been obtained in researches on dielectric phenomena in high-voltage cable insulation and other dielectrics, the various existing theories of dielectric behavior have seemed inadequate. A critical study has therefore been made of these theories in an attempt to obtain a working hypothesis that more nearly meets the stringent requirements of experimental facts. The logarithmic formula is shown to give erroneous results if applied to high-voltage cables when they are operating under high stress. The gradient in a cable must be calculated from the voltampere characteristic of the dielectric when stresses above the elastic limit are used. For stresses below the elastic limit it makes no difference which method is used, but at high stresses an entirely different gradient distribution is obtained when calculated from the volt-ampere characteristic. Likewise, when an insulation is operated above the elastic limit the stress ceases to be a critical factor, but the strain is of utmost importance. In comparing cables that are operating under high voltages, therefore, the strain at the core should be considered rather than the stress at the core. Stress is given by the voltage gradient and strain by the polarization or the current density in the dielectric. Since there is always a conduction current flowing, there must be mobile or free ions present.

This publication has 5 references indexed in Scilit: