Document Type
Thesis
Date of Award
9-30-1989
Degree Name
Master of Science in Electrical Engineering - (M.S.)
Department
Electrical Engineering
First Advisor
Raj Pratap Misra
Second Advisor
S. Pandey
Third Advisor
Mauro Zambuto
Abstract
The main objective of this thesis was to find a correlation between the electrical noise generated within the body of Vulcanized Fibre in insulating service and the time of its breakdown under a synergy of electrical, thermal and mechanical stresses. Failure was defined as a material discontinuity, puncture, under accelerated life test conditions. Radio frequency measurements of the partial discharges occuring in the voids of the material were performed on 1240 samples of vulcanized fibre with 10 mil thickness. The level of electrical noise measured was used to classify the specimens into the three following categories: (a) Low Noise:0-2.8μV (0-9db), (b) Medium Noise:2.8+-5μV (9+-14db) and (c) High Noise: 5+μV (14+db).
Two accelerated life tests were performed. The first involved 90 samples, 30 in each category, at 80°C, 700 VAC and 1.4 psi. During the first test the gathering of complete data indicated a direct correlation of 0.815 by the end of 384 hours. It was noted that by the end of 372 hours only 10% of the low noise samples had failed, 46% of the medium noise samples and 100% of the high noise samples. If the test was discontinued to 372 hours the direct correlation would be 0.989.
The second test involved 60 specimens, 20 in. each category, at 70°C, 700 VAC, 1.4 psi. By gathering complete data a direct correlation of 0.91 was found by the end of 874 hours. It was noted that by the end of 802 hours 100% of the high noise samples had failed, 55% of the medium noise and only 35% of the low noise had failed.
From the above tests a clear correlation can be seen between elecrical noise and time of failure.
It was decided not to subject the data analysis to extrapolation. This decision was based on previous work by Soomro and Sundaresan [2-89]. At their work proved that using Null hypothesis the failure rate between high noise and low noise samples with respect to time had less than one chance in a thousand that the failure could have occured by chance. Therefore the particular failure mode was ignored and only the time at which the breakdowns occured were subjected to statistical analysis using either censored or complete data. With this approach satisfactory results were obtained and one has reasonable ground to support the claim that the electrical noise can be used as a tool to predict the lifetime and the reliability of a given insulating material, since data has been collected for four different dielectric materials up to now.
The other objective of the experimental work involved the collection of data to validate the modified Arrhenious model, proposed by professor R. P. Misra [10-87]. This model can be used to predict life expectancy by data obtained from specimens tested at various constant elevated temperatures. The new parameters introduced in the model take into account the electric field in addition to temperature thus allowing the model to predict life expectancy under combined electrical and thermal stresses. The advantage of this model is that it better simulates the actual stresses encountered by a dielectric under real working conditions.
The experimental work involved the measurement of a large population of vulcanized fibre and the subsequent exposure of groups of samples to elevated temperatures (60°C, 70°C, 80°C). The samples were removed periodically from the enviromental chambers and their change in DC resistance under a test voltage of 1500 VDC was noted. The vulcanised fibre showed an improvement of its dielectric DC resistance well over 500% as the moisture in the material was decreased due to heating. However after 192 hours of continuous heating the dielectric DC resistance dropped to 35% of the original measured value. The samples were exposed to a continuous thermal stress for 516 hours. During this period the decrease of the dielectrics DC resistance was found to be higher for higher temperatures. No indication of eminent failure due to thermal aging was observed. The results obtained were promising but inconclusive. Further research is needed to validate the proposed model whose potential strength will result in the ability to predict with a higher degree of accuracy the reliability of insulating material using electrical noise as a key parameter.
Recommended Citation
Donos, George A., "Reliability prediction of vulcanized fibre using electrical noise as a screening parameter" (1989). Theses. 2751.
https://digitalcommons.njit.edu/theses/2751
