Condition Assessment of Generator Insulation Using Diagnostic Tests Tenzin Chophel* and Thanapong Suwanasri Electrical and Software Systems Engineering

Condition Assessment of Generator Insulation Using Diagnostic Tests

Tenzin Chophel* and Thanapong Suwanasri
Electrical and Software Systems Engineering, The Sirindhorn International Thai-German Graduate School of Engineering (TGGS),
King Mongkut’s University of Technology North Bangkok
1518 Pracharat 1 Rd., Bangsue, Bangkok, Thailand
*[email protected] Cattareya Suwanasri
Department of Electrical and Computer Engineering,
Faculty of Engineering,
King Mongkut’s University of Technology North Bangkok
1518 Pracharat 1 Rd., Bangsue, Bangkok, Thailand
[email protected]

Abstract— Generator is a critical asset in the power system. The reliable operation of a generator mostly depends on the integrity of stator winding insulation system; hence, performing a periodic diagnostic test and finding a health index of generator insulation has become mandatory. In this paper, effort has been made to compute a health index of generator insulation using a scoring and weighting method. Three levels of score are given against each dielectric diagnostic test conducted on a stator winding. The score is given after framing an acceptance criteria referring the relevant standards. The weight of each test is calculated using analytical hierarchy process. Diagnostic test used for this approach includes insulation resistance, polarization index, tan ? (dissipation factor) test, capacitance test, DC ramp test, and partial discharge test. Visual inspection of a stator winding is also included. The test results of 12 in-service hydro generators with F insulation class, commissioned between 1999 and 2006 in Bhutan are applied. After computing a health index, it is concluded that the condition of winding insulation in all 12 hydro generators are remaining satisfactory.

Index terms–winding insulation, diagnostic test, condition assessment, weighting and scoring method, health index.
I. INTRODUCTION
The insulation system of generators is always under influence of one or combination of stresses such as thermal, electrical, mechanical, and ambient which causes the degradation and ageing in the insulation. One of the statistics published by CIGRE study committee 1 states that, 56% of hydro generator failures occurred due to insulation damage as shown in Fig. 1; hence, the health of hydro generator is fairly determined by the condition of insulation in the stator winding.

Fig. 1. Causes of failure in a hydro generator 1.
Thailand International Cooperation Agency (TICA), Thailand and Druk Green Power Corporation Limited (DGPC), Bhutan
The failure in the rotating machine due to insulation breakdown can cause catastrophic damage to the generator as well as prolonged forced outages and huge financial losses to a power utility. By knowing the condition of generators, asset managers can prevent unexpected in-service failure and help in economically planning a maintenance work; hence, diagnostic tests are performed at regular time interval. In scoring and weighting method, a score of “1”, “3”, and “5” are assigned for “Poor”, “Fair” and “Good” conditions respectively against each dielectric diagnostic test conducted on a stator winding. These scores are given as per the acceptance criteria derived from the relevant standards. The dielectric diagnostic tests included in the study are insulation resistance (IR) and polarization index (PI), dissipation factor test (tan ?), capacitance test, DC ramp (DCR) test, and partial discharge (PD) test. A visual inspection (VI) of stator winding is also included.
II. DIELECTRIC DIAGNOSTIC TESTS
A. DC Insulation Resistance Measurement
IR test is important test to get the indication of overall cleanliness, moisture content and general condition of insulation system. The detail of IR measurement and DC voltage to be applied is given in IEEE Std. 43-2013 2 along with the minimum requirement of IR1MIN (IR measured at 1 minute). If IR value decreases over the years, it is the indication of gradual deterioration of insulation. The acceptance criteria derived from above standard and a score assigned is given in Table I.
B. Polarization Index Measurement
PI is the ratio of IR measured at 10 minute divided IR measured at 1 minute (IR10MIN/IR1MIN). In general, a high value of PI indicates the good condition of the insulation system. The acceptance criteria for PI is as also given in Table I.
TABLE I. ACCPETANCE CRITERIA FOR IR AND PI DERIVED FROM IEEE STD 43-2013 AND THEIR SCORES
IR1MIN (G?) PI Score Remark
< 0.1 1.0 > 3 5 Good
C. Partial Discharge Test
Many stator winding failures have PD as a direct cause, or a symptom of the failure process 3. The details of the measurement methods are provided in IEC 60027 4 and IEEE 1434-2000 5. PD is widely measured using a coupling capacitor connected to stator terminal; however, there is no specific limit defined in the standards as the accuracy of PD measurement depends on the ability to correctly identify PD patterns filtering the unwanted external signals as much as possible. Some of the PD experts adopt the pattern recognizing systems such as phase resolved partial discharge (PRPD) analysis whereby the severity of PD is determined through identification of signature/standard patterns as shown in Fig. 2. The severity level of PD is referred as “High”, “Medium” and “Low” depending on the impact of different types of PD on ageing of insulation; accordingly, Table II is the acceptance criteria for the PD types derived from IEC 60027-34-1 standard 6, 7.

Fig. 2. Standard PRPD patterns of high, medium, and low severity in the insulation of generator 7.
TABLE II. ACCEPTANCE CRITERIA AND SCORE FOR PD TEST DERIVED FROM IEC 60027-34-1
PD Types as per PRPD Pattern Score Severity
– Delamination of insulation tapes from the winding conductor ?
– Delamination of insulation tape layers ?, ?
– Abrasion of slot corona protection tape/paint ? 1 High
-End-winding surface discharge/tracking ?, ?, ?, ?
– End-winding discharges in gas/sparking ?
– Bad connection between outer corona protection (OCP) and end potential grading (EPG), ?
– Discharge between OCP and EPG ?, ? 3 Moderate
– Micro voids/cavities ?, ? 5 Low
D. DC Ramp Test
DCR test is a controlled voltage test. It enables a possibility of detecting impending insulation problems by recognizing abnormalities in the measured voltage-current (V-I) response. It also allows to discontinue the test prior to insulation failure. A measurement detail is given in IEEE 95-2002 8. The voltage is smoothly increased at a constant rate usually 1 or 2 kV/minute and V-I response curve is automatically graphed and displayed on the measuring instrument as shown in Fig. 3. The leakage current (Ilea) component of high-quality insulation will usually be small and linear in nature. As the insulation starts to deteriorate, the Ilea will increase. At some voltage level, Ilea will become non-linear or increase drastically over the years if the insulation deterioration is high 8.

Fig. 3. A normal V-I response curve of three phases obtained during DCR test on 11 kV generator U6, Bhutan 2015.
Since Ilea differs from one generator to another, there is no standard limits specified in the standards. In this approach, the V-I response curve is compared among the three phases. The percentage of Ilea deviation (Ilea%dev) calculated by (1) is also compared among the three phases and accordingly, an acceptance criteria is framed as shown in Table III.
(1)
TABLE III. ACCEPTANCE CRITERIA AND SCORE FOR DCR TEST DERIVED FROM IEEE 95-2002
Conditions for DC Ramp Test Score Remark
If the V-I response curve is not similar within the phases and Ilea%dev is greater than %devmean value. 1 Poor
If the V-I response curve is not similar within the phases but Ilea %dev is less than %devmean value OR V-I curve is similar but Ilea %dev is greater than %devmean value. 3 Fair
If the V-I response curve is similar within the phases and Ilea %dev is less than %devmean value 5 Good
E. Tan ? Measurement
Tan ? test is also called a dissipation factor test or a power factor test. Power factor (cos ???is nearly same for a specimen with a tan ? value of 0.100 or less. In an equivalent circuit of a stator winding, the angle between the applied voltage and the voltage dropped across the coil capacitance is called loss angle, ?. There are losses in the winding due to the presence of voids or delamination in the insulation which causes PD consuming some energy. This results in increasing of ? and tan ? 9. So, an increase in tan ? over the years can be an indication of change in the condition insulation. The tan ? is measured at three intervals 0.2Un, 0.6Un and at Un (rated line-to-ground voltage) of the generator as shown Fig. 4. The tan ? tip-up calculated by (2) is also used to study the insulation condition. The tan ? tip-up will increase if there is presence of void or delamination.
Tan ? tip-up = tan ?1.0 – tan ?0.2 (2)
where tan ?0.2 and tan ?1.0 are tan ? generally measured at 0.2Un and at Un respectively.

Fig. 4. Tan ?% vs. test voltage plotted showing the similarity within the phases in the generator of U1, Bhutan, 2015.
In 10, the maximum limits are taken as ?3% and ?1% for tan ? and tan ? tip-up respectively for a phase winding with epoxy mica insulation. In 11, for the same purpose, it is taken as ?6% and ?4% for tan ? and tan ? tip-up respectively; however, there is no standard limit available for a complete stator winding. IEEE 286 12 tells about measurement details and IEC 60034-27-3 9 gives standard limits for individual bar only. Measurement of the tip-up is affected by the presence of silicon carbide stress control coating on coils rated at 6 kV and above as this coating has a non-linear property 3. It acts as a very high resistance coating at low voltage (thus, no power loss in the coating) and when tested at rated voltage, by design the silicon carbide coating will have a relatively low resistance producing more power loss in the coating. The tip-up test is only recommended for trending. For this approach, tan ? and tan ? tip-up values are compared within the three phases for similarity and deviation. An acceptance criteria is set as shown in Table IV
TABLE IV. ACCEPTANCE CRITERIA AND SCORE FOR TAN DELTA TEST DERIVED FROM IEEE 95-2002
Conditions for Tan ? Test Score Remark
– If the %tan ? vs. test voltage curve is not similar within three phases (and)
– If the %dev of %tan ?0.2 and %tan ? tip-up are greater than its %devmean value. 1 Poor
– If the %tan ? vs. test voltage curve is similar within three phases but %dev of %tan ?0.2 and %tan ? tip-up are either equal or not equal to its %devmean value. 3 Fair
-If the %tan ? vs. test voltage curve is similar within three phases (and)
– If the %dev of %tan ?0.2 and %tan ? tip-up is less than its %devmean value. 5 Good
F. Capacitance Measurement
The capacitance (Cap) test can assess the average condition of the insulation. If there is long-time overheating in the winding, there will be gas-filled delamination formed in the groundwall insulation layers. As the percentage of gas within groundwall insulation increases as a result of thermal deterioration, the average dielectric constant will decrease. This is because the filled gas, which is usually air has lower dielectric constant than all solid insulation materials and hence, the capacitance decreases. By trending the change in winding capacitance over the time, a thermal deterioration or problem with moisture or contamination occurring in the winding can be detected 4. If the capacitance tip-up increases over the time, it also indicates the presence of voids or delamination in the insulation and it is calculated by (3).
Capacitance tip-up = (Cap.1.0 – Cap.0.2)/ Cap.0.2 (3)
where Cap.0.2 and Cap.1.0 are capacitance measured at 0.2Un and Un respectively. In 11, the capacitance tip-up limit is taken as ?8%; however, there is no standard limit available. An acceptance criteria for capacitance measurement is set by comparing the capacitance between three phases and calculating the %dev as shown in Table V.
TABLE V. ACCEPTANCE CRITERIA AND SCORE FOR CAPACITANCE MEASUREMENT
Conditions for Capacitance Test Score Remark
– If the capacitance vs. test voltage curve is not similar within all three phases (and)
– If the %dev of Cap0.2 and %Cap tip-up is greater than its %devmean value. 1 Poor
– If the capacitance vs. test voltage curve is similar within three phases but %dev of Cap0.2 and %Cap tip-up is either equal or not equal to its %devmean value. 3 Fair
– If the capacitance vs. test voltage is similar within three phases (and)
– If the %dev of Cap0.2 and %Cap tip-up are less than its %devmean value. 5 Good
G. Visual Inspection of Stator Winding
Visual inspection is one of the most practical and important activities used in determining the condition of stator winding. IEEE 62.2-2004 13 recommends a checklist for carrying out a winding inspection including slot and end-windings of the generator. Accordingly the list of activities in acceptance criteria for visual inspection of the stator winding of generator is derived from this standard and is shown in Table VI.
TABLE VI. ACCEPTANCE CRITERIA AND SCORE FOR VISUAL INSPECTION AS PER THE LIST DERIVED FROM IEEE 62.2-2004
List of Observations Derived from IEEE 62.2-2004 Score
5 (Good) 3 (Fair) 1 (Poor)
Contamination level in the winding Low Moderate High
Small cracks and mechanical damage on the surface of insulation No Moderate High
Overheating or softened insulation No 3 cases
Melted core/wedge No 3 cases
White residue/deposit No No Yes
Corona discharge/treeing No No Yes
Loose bar/coil No No Yes
III. HEALTH INDEX CALCULATION USING SCORING AND WEIGHTING METHOD AND A CASE STUDY
A heath index (HI) calculated by (4) below, is the overall condition of generator winding insulation.
(4)
where Si and Wi are the score and weight respectively against each condition test. Smax is the maximum score, 5. The weight against each test is calculated using Analytical Hierarchy Process (AHP). The calculated weight of each test is shown in Table VII. In the case study, the test results of 12 hydro generators of Druk Green Power Corporation Limited, Bhutan have been analyzed and a HI calculated is shown in Table VIII. The smallest capacity of generator is 12 MW with voltage rating of 11 kV and the biggest is 170 MW, 13.8 kV. These generators were commissioned between the years 1999 to 2006. All generators have class “F” insulation made of epoxy-mica. PD measurement data were not available for some generators; hence, PD severity was considered “moderate” for those generators. The HI of 12 generators are categorized into three levels as shown in Table IX.
TABLE VII. WEIGHT OF EACH DIAGNOSTIC TEST CALCULATED USING AHP
Diagnostic
Tests IR1MIN PI PD DCR Tan ? Cap VI
%Weight (W) 25 25 20 9 5 5 11
TABLE VIII. CALCULATION OF HI OF GENERATOR INSULATION USING WEIGHTING AND SCORING METHOD
Identity of Generator Scores (S) Obtained from 6 Diagnostic Test Results and Visual Inspection (VI) HI %
IR1MIN PI PD DCR Tan ? Cap VI
Unit-1 5 5 3 3 3 3 3 80
Unit-2 5 5 3 5 5 5 3 88
Unit-3 5 5 3 5 5 3 3 86
Unit-4 5 5 3 5 5 5 3 88
Unit-5 5 5 3 3 5 5 3 84
Unit-6 5 5 3 5 5 5 3 88
Unit-7 5 5 3 1 5 5 3 80
Unit-8 5 5 3 1 5 5 3 80
Unit-9 5 3 3 5 5 1 3 74
Unit-10 5 5 3 1 5 1 3 77
Unit-11 5 3 3 5 3 1 3 72
Unit-12 5 5 3 5 3 3 3 84
TABLE IX. HEALTH INDEX AND CONDITION BASED RECOMMENDATIONS
HI % Condition Indicator Condition Based Recommendations
>80 Good Continue operation and maintenance without any restriction
60-80 Acceptable Continue operation but timely re-evaluation is recommended