Design of a Partial Discharge Test Platform author: Daniel Harmsen thesis advisor (mentor): Dr. Armando Rodrigo Mor abstract: Partial discharge (PD) measurements are an effective tool for insulation diagnostics and assessment. Therefore, a good understanding of these PD measurements is an essential part of an electrical engineer’s background knowledge. During this research project a PD test platform was designed and build for electrical PD detection. The setup included artificially created defects for six different types of PDs, with origins in positive and negative corona, internal discharge, floating electrodes, free-moving particles and surface discharge. These defects were designed to have a partial discharge inception voltage (PDIV) of around 10kV, and could easily be connected or disconnected from the setup. Therefore, it was possible to measure individual defects or a combination of them. This PD test platform was used during this research project to characterize the different PD types (single and multiple), and as a check for testing new clustering and pattern recognition techniques. In addition, the platform could also be used as a test platform for educational purposes and to train people and test equipment. In this thesis research project, the “initial step” towards PD defect origin recognition for AC voltage without any phase dependency was conducted with the use of time-resolved partial discharge (TRPD) analysis. This thesis shows that it is possible to conduct the time analysis and recognition for AC voltage for the artificially created defects. From the analysis in this thesis, it can be concluded that the TRPD analysis for PD recognition under AC voltage is as good as, or even better than, phase-resolved partial discharge (PRPD) analysis. For most of the data analysed, the TRPD analysis provides the same results compared to the PRPD analysis. However, further checking is needed, such as validating the findings with mathematical models. The thesis goals are to build a setup suitable for the research, determining the optimal combination of hardware/software to discriminate among different types of defects, and to realize experiments to validate the design.