Linking a physical arc model with a black box arc model and verification

The arc behavior in the current zero region is critical in the case of very steep rising TRV, such as after clearing a short-line fault. Therefore, intensive and abundant short-line fault tests (L90) of a 245 kV SF6 circuit breaker were performed at the KEMA High Power Laboratory. For the purpose of a comparative analysis three different sets of data were obtained during the tests: 1) High-resolution measurements of near current-zero arc current and voltage were carried out. The current zero measurement system (CZM) works as a standalone system in addition to the standard laboratory data acquisition system. The arc conductance shortly before current zero and the arc voltage extinction peak give a clear indication of the interrupting capability of the breaker under test. 2) From the measured traces of every individual test, arc parameters (3 time constants and 3 cooling-power constants) were extracted for the composite black box arc model, which has been developed by KEMA High Power Laboratory and is based on more than 1000 high-resolution measurements during tests of commercial high-voltage circuit breakers. Its aim is to simulate interruption phenomenon in SF6 gas, evaluate performance of HV SF6 circuit breakers in testing and enable the prediction of the performance under conditions other than those tested. 3) After each test, using specially developed computer software, based on a simplified physical enthalpy flow arc model, the values of the arcing contact distance, gas mass flow through the nozzle throat and pressure inside the compression cylinder were calculated. The values of these characteristic quantities at the current zero are relevant indicators for successful interruption. In the comparative analysis, mathematical relations and statistical correlations between the evaluated parameters of the composite black box arc model and the characteristic output quantities are established and discussed. The link has been verified by MatLAB simulation of every individual test. This approach enables acceptable prediction of interruption success in a similar circuit and with a similar interrupter without SLF tests and CZM.