Conference Agenda

Employing of dielectrical losses measurement for insulation of high voltage rotating machines is widespread for the most manufacturers of the winding for high voltage rotating machines. Typically trending over of the dielectrical losses versus losses allows to qualify for extend of the insulation cure and insulation integrity. Losses in the end turns grading system creates significant impact on the measurements and does not allow to use this test without guarding out losses created by the current thought the end turns voltage grading surfaces.

North American standard IEEE 286 and European standard IEC 60034-27-3 recommends different guarding techniques in some extent mitigating of losses due to current though the voltage grading surfaces. Interference to the dielectrical losses measurements depends on guarding technique in different ways and depends also on the basis of the measurements.

Influence of the higher harmonics especially third harmonic both in the test voltage and test set supplied voltage creates additional problem in the testing.

Employing of inner corona protection for single turn coils such as Roebel bars also distort the dielectrical losses measurements.

All these interferences influencing on the routine test of dielectrical losses creating unpredictable distortion of the actual dielectrical losses increasing of the dielectrical losses measured values and sometimes creating negative tip-up of dielectrical losses.

Some specifications inherited from the past restriction on the negative dielectrical losses tip-up. Negative dielectrical losses tip-up was typical in the past in the asphalt matrix micasplitting insulation related. Measured dielectrical losses start to decrease at double of corona inception voltage. Thus, insulation with vast delaminations had dielectrical losses reduction at the higher test voltage. With epoxy or polyester matrix insulation negative dielectrical losses tip-up relates most of the time with distorted measured values.

The present work is attempting to summarize the potential problem with measurements of dielectrical losses using different guarding techniques and different testing devices.

Thermal cycling testing was performed on turbine generator stator bars according to IEEE 1310. Case studies are presented spanning multiple different insulation systems. These case studies encompass varying designs and manufacturing methods (resin rich autoclave, resin rich press cure and vacuum pressure impregnation). These case studies and the results of thermal cycling tests of each are summarized and statistically compared to the performance of non-thermal cycled bars of the same design. The results show that de-bonding of the insulation at the conductor interface does not correlate to a statistically significant change to the dielectric breakdown strength or voltage endurance performance of the thermal cycled bars compared to non-thermal cycled bars. Likewise, a statistical increase of dissipation factor and tip up did not correlate to a statistically significant change in dielectric breakdown strength or voltage endurance performance. The results also demonstrate that the dielectric breakdown strength and voltage endurance tests have a large standard deviation. This indicates that caution must be taken when assessing data from a limited sample size of test and control bars. These case studies challenge the common perception that de-bonding at the conductor interface and the associated increase of dissipation factor after thermal cycling has a significant impact on the stator bar or winding life. These results also reinforce that the critical result of the thermal cycling test is the impact of the test on the dielectric breakdown strength or voltage endurance performance of bars compared to non-thermal cycled bars. This supports that the diagnostic testing described in IEEE 1310 should remain informative as data does not support that these tests correlates to bar or winding life.

The presence of partial discharges (PD) in an insulation system has been a cause for concern in equipment for decades. Some dielectric materials have stronger resistance to degradation from PD than others, hence the many works that discuss partial discharge and its methods of measurement, such as PDIV as well as PD magnitude.

This paper seeks to investigate the affect of partial discharge on dielectric material chemical and mechanical aspects.

Aging theories discussed include parameters from partial discharge activity, such as electric field effects, mechanical degradation as well as chemical aspects caused by corona emission due to exposure of nitric acid, ozone, etc.

The chemical aspect includes components from the degradation of the chemical structure of the dielectric material, such as epoxy, as well as components generated thru interaction with environmental constituents, such as moisture. Additionally the influence on local metallic constituents will be discussed.