Study on Quench Detection of the KSTAR PF Coils Using Numerical Compensation of Inductive Voltages with Genetic Algorithm

A quench detection system for Korea Superconducting Tokamak Advanced Research (KSTAR) Poloidal Field (PF) coils is indispensable against the occurrence of a normal zone due to frequently variable current operation and an enormous amount of stored energy. The co-wound conductive strips and Wheatstone bridges have been adopted in KSTAR to detect abnormal voltage signal caused from quench. Those voltage detection methods have sufficient capability to detect quench occurrence, but additional cancelling out of mutual induction by adjacent coils would improve stability and reliability of quench detection system. A new voltage detection concept, using central difference averaging (CDA) and mutual inductance compensation (MIK) simultaneously, as an advanced technique to compensate the mutual inductive voltage, has been suggested and studied for this study. The CDA and MIK methods were practically adapted to the KSTAR PF 1 coil for the feasibility study. Test results show that the inductance matrix calculated by the FEM tool should be revised because of a revealed discrepancy between measured CDA voltage and calculated MIK voltage. In this study, the inductance matrix was revised by using the 'Genetic algorithm' optimization method to reduce the discrepancy between CDA and MIK effectively. The adoption of the MIK revised inductance matrix was consistent with CDA voltage; mutual inductive voltage was expected to be successfully eliminated.